miga-base 1.2.18.1 → 1.3.0.0

data/utils/FastAAI/fastaai/fastaai

@@ -0,0 +1,4805 @@
+#!/usr/bin/env python3
+
+################################################################################
+"""---0.0 Import Modules---"""
+import subprocess 
+import argparse 
+import datetime 
+import shutil
+import textwrap
+import multiprocessing
+import pickle
+import gzip
+import tempfile
+#Shouldn't play any role.
+#from random import randint
+
+#We could probably remove Path, too.
+#This as well
+import time
+from collections import defaultdict
+import sys
+import os
+from math import floor
+import sqlite3
+#numpy dependency
+import numpy as np
+import io
+import random
+
+import pyrodigal as pd
+import pyhmmer
+
+from collections import namedtuple
+
+from math import ceil
+
+import re
+
+
+class progress_tracker:
+	def __init__(self, total, step_size = 2, message = None, one_line = True):
+		self.current_count = 0
+		self.max_count = total
+		#Book keeping.
+		self.start_time = None
+		self.end_time = None
+		#Show progrexx every [step] percent
+		self.step = step_size
+		self.justify_size = ceil(100/self.step)
+		self.last_percent = 0
+		self.message = message
+		
+		self.pretty_print = one_line
+		
+		self.start()
+
+	def curtime(self):
+		time_format = "%d/%m/%Y %H:%M:%S"
+		timer = datetime.datetime.now()
+		time = timer.strftime(time_format)
+		return time
+		
+	def start(self):
+		print("")
+		if self.message is not None:
+			print(self.message)
+		
+		try:
+			percentage = (self.current_count/self.max_count)*100
+			sys.stdout.write("Completion".rjust(3)+ ' |'+('#'*int(percentage/self.step)).ljust(self.justify_size)+'| ' + ('%.2f'%percentage).rjust(7)+'% ( ' + str(self.current_count) + " of " + str(self.max_count) + ' ) at ' + self.curtime() + "\n")
+			sys.stdout.flush()
+			
+		except:
+			#It's not really a big deal if the progress bar cannot be printed.
+			pass
+	
+	def update(self):
+		self.current_count += 1
+		percentage = (self.current_count/self.max_count)*100
+		try:
+			if percentage // self.step > self.last_percent:
+				if self.pretty_print:
+					sys.stdout.write('\033[A')
+				sys.stdout.write("Completion".rjust(3)+ ' |'+('#'*int(percentage/self.step)).ljust(self.justify_size)+'| ' + ('%.2f'%percentage).rjust(7)+'% ( ' + str(self.current_count) + " of " + str(self.max_count) + ' ) at ' + self.curtime() + "\n")
+				sys.stdout.flush()
+				self.last_percent = percentage // self.step
+			#Bar is always full at the end.
+			if count == self.max_count:
+				if self.pretty_print:
+					sys.stdout.write('\033[A')
+				sys.stdout.write("Completion".rjust(3)+ ' |'+('#'*self.justify_size).ljust(self.justify_size)+'| ' + ('%.2f'%percentage).rjust(7)+'% ( ' + str(self.current_count) + " of " + str(self.max_count) + ' ) at ' + self.curtime() + "\n")
+				sys.stdout.flush()
+				#Add space at end.
+				print("")
+		except:
+			#It's not really a big deal if the progress bar cannot be printed.
+			pass
+					
+					
+#Takes a bytestring from the SQL database and converts it to a numpy array.
+def convert_array(bytestring):
+	return np.frombuffer(bytestring, dtype = np.int32)
+
+def convert_float_array_16(bytestring):
+	return np.frombuffer(bytestring, dtype = np.float16)
+
+def convert_float_array_32(bytestring):
+	return np.frombuffer(bytestring, dtype = np.float32)
+	
+def convert_float_array_64(bytestring):
+	return np.frombuffer(bytestring, dtype = np.float64)
+
+def read_fasta(file):
+	cur_seq = ""
+	cur_prot = ""
+	
+	contents = {}
+	deflines = {}
+	
+	fasta = agnostic_reader(file)
+	for line in fasta:
+		if line.startswith(">"):
+			if len(cur_seq) > 0:
+				contents[cur_prot] = cur_seq
+				deflines[cur_prot] = defline
+				
+			cur_seq = ""
+			cur_prot = line.strip().split()[0][1:]
+			defline = line.strip()[len(cur_prot)+1 :].strip()
+			
+		else:
+			cur_seq += line.strip()
+				
+	fasta.close()
+	
+	#Final iter
+	if len(cur_seq) > 0:
+		contents[cur_prot] = cur_seq
+		deflines[cur_prot] = defline
+		
+	return contents, deflines
+	
+class fasta_file:
+	def __init__(self, file, type = "genome"):
+		self.file_path = os.path.abspath(file)
+		self.name = os.path.basename(file)
+		self.no_ext = os.path.splitext(self.name)[0]
+		self.type = type
+		
+		self.tuple_structure = namedtuple("fasta", ["seqid", "description", "sequence"])
+		self.contents = {}
+		
+	def convert(self, contents, descriptions):
+		for protein in contents:
+			self.contents = self.tuple_structure(seqid = protein, description = descriptions[protein], sequence = contents[protein])
+			
+	
+	def def_import_file(self):
+		contents, descriptions = read_fasta(self.file_path)
+		self.convert(contents, descriptions)
+		
+class pyhmmer_manager:
+	def __init__(self, do_compress):
+		self.hmm_model = []
+		self.hmm_model_optimized = None
+		
+		self.proteins_to_search = []
+		self.protein_descriptions = None
+		
+		self.hmm_result_proteins = []
+		self.hmm_result_accessions = []
+		self.hmm_result_scores = []
+		
+		self.printable_lines = []
+		
+		self.bacterial_SCPs = None
+		self.archaeal_SCPs = None
+		self.assign_hmm_sets()
+		self.domain_counts = {"Bacteria" : 0, "Archaea": 0}
+		self.voted_domain = {"Bacteria" : len(self.bacterial_SCPs), "Archaea" : len(self.archaeal_SCPs)}
+		
+		self.bacterial_fraction = None
+		self.archaeal_fraction = None
+		
+		self.best_hits = None
+		
+		self.do_compress = do_compress
+		
+	def optimize_models(self):
+		try:
+			self.hmm_model_optimized = []
+			
+			for hmm in self.hmm_model:
+				prof = pyhmmer.plan7.Profile(M = hmm.insert_emissions.shape[0], alphabet = pyhmmer.easel.Alphabet.amino())
+				prof.configure(hmm = hmm, background = pyhmmer.plan7.Background(alphabet = pyhmmer.easel.Alphabet.amino()), L = hmm.insert_emissions.shape[0]-1)
+				optim = prof.optimized()
+				self.hmm_model_optimized.append(optim)
+				
+			#Clean up.
+			self.hmm_model = None
+		except:
+			#Quiet fail condition - fall back on default model.
+			self.hmm_model_optimized = None
+		
+	#Load HMM and try to optimize.
+	def load_hmm_from_file(self, hmm_path):
+		hmm_set = pyhmmer.plan7.HMMFile(hmm_path)
+		for hmm in hmm_set:
+			self.hmm_model.append(hmm)
+			
+		#This doesn't seem to be improving performance currently.
+		self.optimize_models()
+			
+	#Set archaeal and bacterial HMM sets.
+	def assign_hmm_sets(self):
+		self.bacterial_SCPs = {'PF00709_21': 'Adenylsucc_synt', 'PF00406_22': 'ADK', 'PF01808_18': 'AICARFT_IMPCHas', 'PF00231_19': 'ATP-synt',
+		'PF00119_20': 'ATP-synt_A', 'PF01264_21': 'Chorismate_synt', 'PF00889_19': 'EF_TS', 'PF01176_19': 'eIF-1a',
+		'PF02601_15': 'Exonuc_VII_L', 'PF01025_19': 'GrpE', 'PF01725_16': 'Ham1p_like', 'PF01715_17': 'IPPT',
+		'PF00213_18': 'OSCP', 'PF01195_19': 'Pept_tRNA_hydro', 'PF00162_19': 'PGK', 'PF02033_18': 'RBFA', 'PF02565_15': 'RecO_C',
+		'PF00825_18': 'Ribonuclease_P', 'PF00687_21': 'Ribosomal_L1', 'PF00572_18': 'Ribosomal_L13',
+		'PF00238_19': 'Ribosomal_L14', 'PF00252_18': 'Ribosomal_L16', 'PF01196_19': 'Ribosomal_L17',
+		'PF00861_22': 'Ribosomal_L18p', 'PF01245_20': 'Ribosomal_L19', 'PF00453_18': 'Ribosomal_L20',
+		'PF00829_21': 'Ribosomal_L21p', 'PF00237_19': 'Ribosomal_L22', 'PF00276_20': 'Ribosomal_L23',
+		'PF17136_4': 'ribosomal_L24', 'PF00189_20': 'Ribosomal_S3_C', 'PF00281_19': 'Ribosomal_L5', 'PF00181_23': 'Ribosomal_L2',
+		'PF01016_19': 'Ribosomal_L27', 'PF00828_19': 'Ribosomal_L27A', 'PF00830_19': 'Ribosomal_L28',
+		'PF00831_23': 'Ribosomal_L29', 'PF00297_22': 'Ribosomal_L3', 'PF01783_23': 'Ribosomal_L32p',
+		'PF01632_19': 'Ribosomal_L35p', 'PF00573_22': 'Ribosomal_L4', 'PF00347_23': 'Ribosomal_L6',
+		'PF03948_14': 'Ribosomal_L9_C', 'PF00338_22': 'Ribosomal_S10', 'PF00411_19': 'Ribosomal_S11',
+		'PF00416_22': 'Ribosomal_S13', 'PF00312_22': 'Ribosomal_S15', 'PF00886_19': 'Ribosomal_S16',
+		'PF00366_20': 'Ribosomal_S17', 'PF00203_21': 'Ribosomal_S19', 'PF00318_20': 'Ribosomal_S2',
+		'PF01649_18': 'Ribosomal_S20p', 'PF01250_17': 'Ribosomal_S6', 'PF00177_21': 'Ribosomal_S7',
+		'PF00410_19': 'Ribosomal_S8', 'PF00380_19': 'Ribosomal_S9', 'PF00164_25': 'Ribosom_S12_S23',
+		'PF01193_24': 'RNA_pol_L', 'PF01192_22': 'RNA_pol_Rpb6', 'PF01765_19': 'RRF', 'PF02410_15': 'RsfS',
+		'PF03652_15': 'RuvX', 'PF00584_20': 'SecE', 'PF03840_14': 'SecG', 'PF00344_20': 'SecY', 'PF01668_18': 'SmpB',
+		'PF00750_19': 'tRNA-synt_1d', 'PF01746_21': 'tRNA_m1G_MT', 'PF02367_17': 'TsaE', 'PF02130_17': 'UPF0054',
+		'PF02699_15': 'YajC'}
+		
+		self.archaeal_SCPs = {'PF00709_21': 'Adenylsucc_synt', 'PF05221_17': 'AdoHcyase', 'PF01951_16': 'Archease', 'PF01813_17': 'ATP-synt_D',
+		'PF01990_17': 'ATP-synt_F', 'PF01864_17': 'CarS-like', 'PF01982_16': 'CTP-dep_RFKase', 'PF01866_17': 'Diphthamide_syn',
+		'PF04104_14': 'DNA_primase_lrg', 'PF01984_20': 'dsDNA_bind', 'PF04010_13': 'DUF357', 'PF04019_12': 'DUF359',
+		'PF04919_12': 'DUF655', 'PF01912_18': 'eIF-6', 'PF05833_11': 'FbpA', 'PF01725_16': 'Ham1p_like',
+		'PF00368_18': 'HMG-CoA_red', 'PF00334_19': 'NDK', 'PF02006_16': 'PPS_PS', 'PF02996_17': 'Prefoldin',
+		'PF01981_16': 'PTH2', 'PF01948_18': 'PyrI', 'PF00687_21': 'Ribosomal_L1', 'PF00572_18': 'Ribosomal_L13',
+		'PF00238_19': 'Ribosomal_L14', 'PF00827_17': 'Ribosomal_L15e', 'PF00252_18': 'Ribosomal_L16',
+		'PF01157_18': 'Ribosomal_L21e', 'PF00237_19': 'Ribosomal_L22', 'PF00276_20': 'Ribosomal_L23',
+		'PF16906_5': 'Ribosomal_L26', 'PF00831_23': 'Ribosomal_L29', 'PF00297_22': 'Ribosomal_L3',
+		'PF01198_19': 'Ribosomal_L31e', 'PF01655_18': 'Ribosomal_L32e', 'PF01780_19': 'Ribosomal_L37ae',
+		'PF00832_20': 'Ribosomal_L39', 'PF00573_22': 'Ribosomal_L4', 'PF00935_19': 'Ribosomal_L44', 'PF17144_4': 'Ribosomal_L5e',
+		'PF00347_23': 'Ribosomal_L6', 'PF00411_19': 'Ribosomal_S11', 'PF00416_22': 'Ribosomal_S13',
+		'PF00312_22': 'Ribosomal_S15', 'PF00366_20': 'Ribosomal_S17', 'PF00833_18': 'Ribosomal_S17e',
+		'PF00203_21': 'Ribosomal_S19', 'PF01090_19': 'Ribosomal_S19e', 'PF00318_20': 'Ribosomal_S2',
+		'PF01282_19': 'Ribosomal_S24e', 'PF01667_17': 'Ribosomal_S27e', 'PF01200_18': 'Ribosomal_S28e',
+		'PF01015_18': 'Ribosomal_S3Ae', 'PF00177_21': 'Ribosomal_S7', 'PF00410_19': 'Ribosomal_S8',
+		'PF01201_22': 'Ribosomal_S8e', 'PF00380_19': 'Ribosomal_S9', 'PF00164_25': 'Ribosom_S12_S23',
+		'PF06026_14': 'Rib_5-P_isom_A', 'PF01351_18': 'RNase_HII', 'PF13656_6': 'RNA_pol_L_2',
+		'PF01194_17': 'RNA_pol_N', 'PF03874_16': 'RNA_pol_Rpb4', 'PF01192_22': 'RNA_pol_Rpb6',
+		'PF01139_17': 'RtcB', 'PF00344_20': 'SecY', 'PF06093_13': 'Spt4', 'PF00121_18': 'TIM', 'PF01994_16': 'Trm56',
+		'PF00749_21': 'tRNA-synt_1c', 'PF00750_19': 'tRNA-synt_1d', 'PF13393_6': 'tRNA-synt_His',
+		'PF01142_18': 'TruD', 'PF01992_16': 'vATP-synt_AC39', 'PF01991_18': 'vATP-synt_E', 'PF01496_19': 'V_ATPase_I'}
+		
+	#Convert passed sequences.
+	def convert_protein_seqs_in_mem(self, contents):
+		#Clean up.
+		self.proteins_to_search = []
+		
+		for protein in contents:
+			#Skip a protein if it's longer than 100k AA.
+			if len(contents[protein]) >= 100000:
+				continue
+			as_bytes = protein.encode()
+			#Pyhmmer digitization of sequences for searching.
+			easel_seq = pyhmmer.easel.TextSequence(name = as_bytes, sequence = contents[protein])
+			easel_seq = easel_seq.digitize(pyhmmer.easel.Alphabet.amino())
+			self.proteins_to_search.append(easel_seq)
+			
+		easel_seq = None		
+			
+	def load_protein_seqs_from_file(self, prots_file):
+		#Pyhmmer has a method for loading a fasta file, but we need to support gzipped inputs, so we do it manually.
+		contents, deflines = read_fasta(prots_file)
+		self.protein_descriptions = deflines
+		self.convert_protein_seqs_in_mem(contents)
+		
+	def execute_search(self):
+		if self.hmm_model_optimized is None:
+			top_hits = list(pyhmmer.hmmsearch(self.hmm_model, self.proteins_to_search, cpus=1, bit_cutoffs="trusted"))
+		else:
+			top_hits = list(pyhmmer.hmmsearch(self.hmm_model_optimized, self.proteins_to_search, cpus=1, bit_cutoffs="trusted"))
+		
+		self.printable_lines = []
+		
+		self.hmm_result_proteins = []
+		self.hmm_result_accessions = []
+		self.hmm_result_scores = []
+		
+		for model in top_hits:
+			for hit in model:
+				target_name = hit.name.decode()
+				target_acc = hit.accession
+				if target_acc is None:
+					target_acc = "-"
+				else:
+					target_acc = target_acc.decode()
+				
+				query_name = hit.best_domain.alignment.hmm_name.decode()
+				query_acc = hit.best_domain.alignment.hmm_accession.decode()
+				
+				full_seq_evalue = "%.2g" % hit.evalue
+				full_seq_score = round(hit.score, 1)
+				full_seq_bias = round(hit.bias, 1)
+				
+				best_dom_evalue = "%.2g" % hit.best_domain.alignment.domain.i_evalue
+				best_dom_score = round(hit.best_domain.alignment.domain.score, 1)
+				best_dom_bias = round(hit.best_domain.alignment.domain.bias, 1)
+
+				#I don't know how to get most of these values.
+				exp = 0
+				reg = 0
+				clu = 0
+				ov  = 0
+				env = 0
+				dom = len(hit.domains)
+				rep = 0
+				inc = 0
+				
+				try:
+					description = self.protein_descriptions[target_name]
+				except:
+					description = ""
+				
+				writeout = [target_name, target_acc, query_name, query_acc, full_seq_evalue, \
+				full_seq_score, full_seq_bias, best_dom_evalue, best_dom_score, best_dom_bias, \
+				exp, reg, clu, ov, env, dom, rep, inc, description]
+				
+				#Format and join.
+				writeout = [str(i) for i in writeout]
+				writeout = '\t'.join(writeout)
+				
+				self.printable_lines.append(writeout)
+				
+				self.hmm_result_proteins.append(target_name)
+				self.hmm_result_accessions.append(query_acc)
+				self.hmm_result_scores.append(best_dom_score)
+		
+	def filter_to_best_hits(self):
+		hmm_file = np.transpose(np.array([self.hmm_result_proteins, self.hmm_result_accessions, self.hmm_result_scores]))
+		
+		#hmm_file = np.loadtxt(hmm_file_name, comments = '#', usecols = (0, 3, 8), dtype=(str))
+		#Sort the hmm file based on the score column in descending order.
+		hmm_file = hmm_file[hmm_file[:,2].astype(float).argsort()[::-1]]
+		
+		#Identify the first row where each gene name appears, after sorting by score; 
+		#in effect, return the highest scoring assignment per gene name
+		#Sort the indices of the result to match the score-sorted table instead of alphabetical order of gene names
+		hmm_file = hmm_file[np.sort(np.unique(hmm_file[:,0], return_index = True)[1])]
+		
+		#Filter the file again for the unique ACCESSION names, since we're only allowed one gene per accession, I guess?
+		#Don't sort the indices, we don't care about the scores anymore.
+		hmm_file = hmm_file[np.unique(hmm_file[:,1], return_index = True)[1]]
+		
+		sql_friendly_names = [i.replace(".", "_") for i in hmm_file[:,1]]
+		
+		self.best_hits = dict(zip(hmm_file[:,0], sql_friendly_names))
+		
+		hmm_file = None
+	
+	#Count per-dom occurs.
+	def assign_domain(self):
+		for prot in self.best_hits.values():
+			if prot in self.bacterial_SCPs:
+				self.domain_counts["Bacteria"] += 1
+			if prot in self.archaeal_SCPs:
+				self.domain_counts["Archaea"] += 1
+			
+		self.bacterial_fraction = self.domain_counts["Bacteria"] / self.voted_domain["Bacteria"]
+		self.aechaeal_fraction = self.domain_counts["Archaea"]  / self.voted_domain["Archaea"]
+			
+		if self.bacterial_fraction >= self.aechaeal_fraction:
+			self.voted_domain = "Bacteria"
+		else:
+			self.voted_domain = "Archaea"
+			
+		pop_keys = list(self.best_hits.keys())
+		for key in pop_keys:
+			if self.voted_domain == "Bacteria":
+				if self.best_hits[key] not in self.bacterial_SCPs:
+					self.best_hits.pop(key)
+			if self.voted_domain == "Archaea":
+				if self.best_hits[key] not in self.archaeal_SCPs:
+					self.best_hits.pop(key)
+				
+	def to_hmm_file(self, output):
+		#PyHMMER data is a bit hard to parse. For each result:
+		
+		content = '\n'.join(self.printable_lines) + '\n'
+		
+		if self.do_compress:
+			#Clean
+			if os.path.exists(output):
+				os.remove(output)
+				
+			content = content.encode()
+						
+			fh = gzip.open(output+".gz", "wb")
+			fh.write(content)
+			fh.close()
+			content = None
+			
+		else:
+			#Clean
+			if os.path.exists(output+".gz"):
+				os.remove(output+".gz")
+				
+			fh = open(output, "w")
+			
+			fh.write(content)
+				
+			fh.close()
+			
+		content = None
+		
+	#If we're doing this step at all, we've either loaded the seqs into mem by reading the prot file
+	#or have them in mem thanks to pyrodigal.
+	def run_for_fastaai(self, prots, hmm_output):
+		try:
+			self.convert_protein_seqs_in_mem(prots)
+			self.execute_search()
+			self.filter_to_best_hits()
+			try:
+				self.to_hmm_file(hmm_output)
+			except:
+				print(output, "cannot be created. HMM search failed. This file will be skipped.")
+
+		except:
+			print(output, "failed to run through HMMER!")
+			self.best_hits = None
+
+		
+def hmm_preproc_initializer(hmm_file, do_compress = False):
+	global hmm_manager
+	hmm_manager = pyhmmer_manager(do_compress)
+	hmm_manager.load_hmm_from_file(hmm_file)
+	
+class pyrodigal_manager:
+	def __init__(self, file = None, aa_out = None, nt_out = None, is_meta = False, full_headers = True, trans_table = 11,
+				num_bp_fmt = True, verbose = True, do_compress = "0", compare_against = None):
+		#Input NT sequences
+		self.file = file
+		
+		#List of seqs read from input file.
+		self.sequences = None
+		#Concatenation of up to first 32 million bp in self.sequences - prodigal caps at this point.
+		self.training_seq = None
+		
+		#Predicted genes go here
+		self.predicted_genes = None
+		#Record the translation table used.
+		self.trans_table = trans_table
+		
+		#This is the pyrodigal manager - this does the gene predicting.
+		self.manager = pd.OrfFinder(meta=is_meta)
+		self.is_meta = is_meta
+		
+		#Full prodigal header information includes more than just a protein number.
+		#If full_headers is true, protein deflines will match prodigal; else, just protein ID.
+		self.full_headers = full_headers
+		
+		#Prodigal prints info to console. I enhanced the info and made printing default, but also allow them to be totally turned off.
+		self.verbose = verbose
+		
+		#Prodigal formats outputs with 70 bases per line max
+		self.num_bp_fmt = num_bp_fmt
+		
+		#File names for outputs
+		self.aa_out = aa_out
+		self.nt_out = nt_out
+		
+		#List of proteins in excess of 100K base pairs (HMMER's limit) and their lengths. This is also fastAAI specific.
+		self.excluded_seqs = {}
+		
+		#Gzip outputs if asked.
+		self.compress = do_compress
+		
+		self.labeled_proteins = None
+		
+		#Normally, we don't need to keep an input sequence after it's had proteins predicted for it - however
+		#For FastAAI and MiGA's purposes, comparisons of two translation tables is necessary.
+		#Rather than re-importing sequences and reconstructing the training sequences, 
+		#keep them for faster repredict with less I/O
+		self.compare_to = compare_against
+		if self.compare_to is not None:
+			self.keep_seqs = True
+			self.keep_after_train = True
+		else:
+			self.keep_seqs = False
+			self.keep_after_train = False
+	
+	#Imports a fasta as binary.
+	def import_sequences(self):
+		if self.sequences is None:
+			self.sequences = {}
+			
+		#check for zipped and import as needed.
+		with open(self.file, 'rb') as test_gz:
+			#Gzip magic number
+			is_gz = (test_gz.read(2) == b'\x1f\x8b')
+		
+		if is_gz:
+			fh = gzip.open(self.file)
+		else:
+			fh = open(self.file, "rb")
+		
+		imp = fh.readlines()
+		
+		fh.close()
+		
+		cur_seq = None
+		for s in imp:
+			s = s.decode().strip()
+			#> is 62 in ascii. This is asking if the first character is '>'
+			if s.startswith(">"):
+				#Skip first cycle, then do for each after
+				if cur_seq is not None:
+					self.sequences[cur_seq] = ''.join(self.sequences[cur_seq])
+					self.sequences[cur_seq] = self.sequences[cur_seq].encode()
+					#print(cur_seq, len(self.sequences[cur_seq]))
+				cur_seq = s[1:]
+				cur_seq = cur_seq.split()[0]
+				cur_seq = cur_seq.encode('utf-8')
+				self.sequences[cur_seq] = []
+			else:
+				#Remove the newline character.
+				#bases = s[:-1]
+				self.sequences[cur_seq].append(s)
+		
+		#Final set
+		self.sequences[cur_seq] = ''.join(self.sequences[cur_seq])
+		self.sequences[cur_seq] = self.sequences[cur_seq].encode()
+		
+		#Now we have the data, go to training.
+		if not self.is_meta:
+			self.train_manager()
+		
+	#Collect up to the first 32 million bases for use in training seq.
+	def train_manager(self):
+		running_sum = 0
+		seqs_added = 0
+		if self.training_seq is None:
+			self.training_seq = []
+			for seq in self.sequences:
+				running_sum += len(self.sequences[seq])
+				if seqs_added > 0:
+					#Prodigal interleaving logic - add this breaker between sequences, starting at sequence 2
+					self.training_seq.append(b'TTAATTAATTAA')
+					running_sum += 12
+					
+				seqs_added += 1
+					
+				#Handle excessive size
+				if running_sum >= 32000000:					
+					print("Warning:  Sequence is long (max 32000000 for training).")
+					print("Training on the first 32000000 bases.")
+				
+					to_remove = running_sum - 32000000
+					
+					#Remove excess characters
+					cut_seq = self.sequences[seq][:-to_remove]
+					#Add the partial seq
+					self.training_seq.append(cut_seq)
+					
+					#Stop the loop and move to training
+					break
+				
+				#add in a full sequence
+				self.training_seq.append(self.sequences[seq])
+
+			if seqs_added > 1:
+				self.training_seq.append(b'TTAATTAATTAA')
+				
+			self.training_seq = b''.join(self.training_seq)
+		
+		if len(self.training_seq) < 20000:
+			if self.verbose:
+				print("Can't train on 20 thousand or fewer characters. Switching to meta mode.")
+			self.manager = pd.OrfFinder(meta=True)
+			self.is_meta = True
+		else:
+			if self.verbose:
+				print("")
+				#G is 71, C is 67; we're counting G + C and dividing by the total.
+				gc = round(((self.training_seq.count(67) + self.training_seq.count(71))/ len(self.training_seq)) * 100, 2)
+				print(len(self.training_seq), "bp seq created,", gc, "pct GC")
+				
+			#Train
+			self.manager.train(self.training_seq, translation_table = self.trans_table)
+		
+		if not self.keep_after_train:
+			#Clean up
+			self.training_seq = None
+		
+	def predict_genes(self):
+		if self.is_meta:
+			if self.verbose:
+				print("Finding genes in metagenomic mode")
+		else:
+			if self.verbose:
+				print("Finding genes with translation table", self.trans_table)
+				print("")
+			
+		self.predicted_genes = {}
+		for seq in self.sequences:
+			
+			if self.verbose:
+				print("Finding genes in sequence", seq.decode(), "("+str(len(self.sequences[seq]))+ " bp)... ", end = '')
+				
+			self.predicted_genes[seq] = self.manager.find_genes(self.sequences[seq])
+				
+			#If we're comparing multiple tables, then we want to keep these for re-prediction.
+			if not self.keep_seqs:
+				#Clean up
+				self.sequences[seq] = None
+			
+			if self.verbose:
+				print("done!")
+			
+	#Predict genes with an alternative table, compare results, and keep the winner.	
+	def compare_alternative_table(self, table):
+		if table == self.trans_table:
+			print("You're trying to compare table", table, "with itself.")
+		else:
+			if self.verbose:
+				print("Comparing translation table", self.trans_table, "against table", table)
+			old_table = self.trans_table
+			old_genes = self.predicted_genes
+			old_size = 0
+			for seq in self.predicted_genes:
+				for gene in self.predicted_genes[seq]:
+					old_size += (gene.end - gene.begin)
+			
+			self.trans_table = table
+			self.train_manager()
+			self.predict_genes()
+				
+			new_size = 0
+			for seq in self.predicted_genes:
+				for gene in self.predicted_genes[seq]:
+					new_size += (gene.end - gene.begin)
+			
+			if (old_size / new_size) > 1.1:
+				if self.verbose:
+					print("Translation table", self.trans_table, "performed better than table", old_table, "and will be used instead.")
+			else:
+				if self.verbose:
+					print("Translation table", self.trans_table, "did not perform significantly better than table", old_table, "and will not be used.")
+				self.trans_table = old_table
+				self.predicted_genes = old_genes
+			
+			#cleanup
+			old_table = None
+			old_genes = None
+			old_size = None
+			new_size = None
+		
+	def predict_and_compare(self):
+		self.predict_genes()
+	
+		#Run alt comparisons in gene predict.
+		if self.compare_to is not None:
+			while len(self.compare_to) > 0:
+				try:
+					next_table = int(self.compare_to.pop(0))
+					
+					if len(self.compare_to) == 0:
+						#Ready to clean up.
+						self.keep_after_train = True
+						self.keep_seqs = True
+					
+					self.compare_alternative_table(next_table)
+				except:
+					print("Alternative table comparison failed! Skipping.")
+		
+	#Break lines into size base pairs per line. Prodigal's default for bp is 70, aa is 60.
+	def num_bp_line_format(self, string, size = 70):
+		#ceiling funciton without the math module
+		ceiling = int(round((len(string)/size)+0.5, 0))
+		formatted = '\n'.join([string[(i*size):(i+1)*size] for i in range(0, ceiling)])
+		return formatted
+	
+	#Writeouts
+	def write_nt(self):
+		if self.nt_out is not None:
+			if self.verbose:
+				print("Writing nucleotide sequences... ")
+			if self.compress == '1' or self.compress == '2':
+				out_writer = gzip.open(self.nt_out+".gz", "wb")
+				
+				content = b''
+				
+				for seq in self.predicted_genes:
+					seqname = b">"+ seq + b"_"
+					#Gene counter
+					count = 1
+					for gene in self.predicted_genes[seq]:
+						#Full header lines
+						if self.full_headers:
+							content += b' # '.join([seqname + str(count).encode(), str(gene.begin).encode(), str(gene.end).encode(), str(gene.strand).encode(), gene._gene_data.encode()])
+						else:
+							#Reduced headers if we don't care.
+							content += seqname + str(count).encode()
+							
+						content += b'\n'
+							
+						if self.num_bp_fmt:
+							#60 bp cap per line
+							content += self.num_bp_line_format(gene.sequence(), size = 70).encode()
+						else:
+							#One-line sequence.
+							content += gene.sequence().encode()
+							
+						content += b'\n'
+						count += 1
+				
+				out_writer.write(content)
+				out_writer.close()
+			
+			if self.compress == '0' or self.compress == '2':
+				out_writer = open(self.nt_out, "w")
+			
+				for seq in self.predicted_genes:
+					#Only do this decode once.
+					seqname = ">"+ seq.decode() +"_"
+					#Gene counter
+					count = 1
+					
+					for gene in self.predicted_genes[seq]:
+						#Full header lines
+						if self.full_headers:
+							#Standard prodigal header
+							print(seqname + str(count), gene.begin, gene.end, gene.strand, gene._gene_data, sep = " # ", file = out_writer)
+						else:
+							#Reduced headers if we don't care.
+							print(seqname + str(count), file = out_writer)
+							
+						if self.num_bp_fmt:
+							#60 bp cap per line
+							print(self.num_bp_line_format(gene.sequence(), size = 70), file = out_writer)
+						else:
+							#One-line sequence.
+							print(gene.sequence(), file = out_writer)
+							
+						count += 1
+							
+				out_writer.close()
+		
+	def write_aa(self):
+		if self.aa_out is not None:
+			if self.verbose:
+				print("Writing amino acid sequences...")
+				
+			self.labeled_proteins = {}
+			content = ''
+			for seq in self.predicted_genes:
+				count = 1
+				seqname = ">"+ seq.decode() + "_"
+				for gene in self.predicted_genes[seq]:
+					prot_name = seqname + str(count)
+					translation = gene.translate()
+					self.labeled_proteins[prot_name[1:]] = translation
+					defline = " # ".join([prot_name, str(gene.begin), str(gene.end), str(gene.strand), str(gene._gene_data)])
+					content += defline
+					content += "\n"
+					count += 1
+					content += self.num_bp_line_format(translation, size = 60)
+					content += "\n"					
+				
+			if self.compress == '0' or self.compress == '2':
+				out_writer = open(self.aa_out, "w")
+				out_writer.write(content)
+				out_writer.close()
+				
+			if self.compress == '1' or self.compress == '2':
+				content = content.encode()
+				out_writer = gzip.open(self.aa_out+".gz", "wb")
+				out_writer.write(content)
+				out_writer.close()
+				
+	def run_for_fastaai(self):
+		self.verbose = False
+		self.import_sequences()
+		self.train_manager()
+		self.predict_and_compare()
+		self.write_aa()
+	
+#Iterator for agnostic reader
+class agnostic_reader_iterator:
+	def __init__(self, reader):
+		self.handle_ = reader.handle
+		self.is_gz_ = reader.is_gz
+		
+	def __next__(self):
+		if self.is_gz_:
+			line = self.handle_.readline().decode()
+		else:
+			line = self.handle_.readline()
+		
+		#Ezpz EOF check
+		if line:
+			return line
+		else:
+			raise StopIteration
+
+#File reader that doesn't care if you give it a gzipped file or not.
+class agnostic_reader:
+	def __init__(self, file):
+		self.path = file
+		
+		with open(file, 'rb') as test_gz:
+			#Gzip magic number
+			is_gz = (test_gz.read(2) == b'\x1f\x8b')
+		
+		self.is_gz = is_gz
+		
+		if is_gz:
+			self.handle = gzip.open(self.path)
+		else:
+			self.handle = open(self.path)
+			
+	def __iter__(self):
+		return agnostic_reader_iterator(self)
+		
+	def close(self):
+		self.handle.close()
+
+'''
+Class for handling all of the raw genome/protein/protein+HMM file inputs when building a database.
+
+Takes a file or files and processes them from genome -> protein, protein -> hmm, prot+HMM -> kmerized protein best hits as numpy int arrays according to the kmer_index
+
+'''
+
+class input_file:
+	def __init__(self, input_path, output = "", verbosity = False, do_compress = False, 
+	make_crystal = False):
+		#starting path for the file; irrelevant for protein and hmm, but otherwise useful for keeping track.
+		self.path = input_path
+		#Output directory starts with this
+		self.output = os.path.normpath(output + "/")
+		#For printing file updates, this is the input name
+		self.name = os.path.basename(input_path)
+		#original name is the key used for the genomes index later on.
+		self.original_name = os.path.basename(input_path)
+		#This is the name that can be used for building files with new extensions.
+		if input_path.endswith(".gz"):
+			#Remove .gz first to make names consistent.
+			self.basename = os.path.splitext(os.path.basename(input_path[:-3]))[0]
+		else:
+			self.basename = os.path.splitext(os.path.basename(input_path))[0]
+			
+		#Sanitize for SQL
+		#These are chars safe for sql
+		sql_safe = set('_abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789')
+		current_chars = set(self.basename)
+		#self.sql_name = self.basename
+		#Identify SQL-unsafe characters as those outside the permissible set and replace all with underscores.
+		for char in current_chars - sql_safe:
+			self.basename = self.basename.replace(char, "_")
+			
+		#'genome' or 'protein' or 'protein and HMM' 
+		self.status = None
+		#These will keep track of paths for each stage of file for us.
+		self.genome = None
+		self.protein = None
+		self.hmm = None
+		
+		self.ran_hmmer = False
+		
+		#If pyrodigal is run, then the protein sequences are already loaded into memory. 
+		#We reuse them in kmer extraction instead of another I/O
+		self.prepared_proteins = None
+		
+		self.intermediate = None
+		
+		self.crystalize = make_crystal
+		self.best_hits = None
+		self.best_hits_kmers = None
+		
+		self.protein_count = 0
+		self.protein_kmer_count = {}
+		
+		self.trans_table = None
+		self.start_time = None
+		self.end_time = None
+		self.err_log = ""
+		#doesn't get updated otw.
+		self.initial_state = "protein+HMM"
+		
+		self.verbose = verbosity
+		
+		#Check if the file failed to produce ANY SCP HMM hits.
+		self.is_empty = False
+		
+		self.do_compress = do_compress
+	
+		self.crystal = None
+		
+		self.init_time = None
+		#default to 0 time.
+		self.prot_pred_time = None
+		self.hmm_search_time = None
+		self.besthits_time = None
+	
+	def curtime(self):
+		time_format = "%d/%m/%Y %H:%M:%S"
+		timer = datetime.datetime.now()
+		time = timer.strftime(time_format)
+		return time
+		
+	def partial_timings(self):
+		protein_pred = self.prot_pred_time-self.init_time
+		hmm_search = self.hmm_search_time-self.prot_pred_time
+		besthits = self.besthits_time-self.hmm_search_time
+		
+		protein_pred = protein_pred.total_seconds()
+		hmm_search = hmm_search.total_seconds()
+		besthits = besthits.total_seconds()
+		
+		self.prot_pred_time = protein_pred
+		self.hmm_search_time = hmm_search
+		self.besthits_time = besthits
+	
+	#Functions for externally setting status and file paths of particular types
+	def set_genome(self, path):
+		self.status = 'genome'
+		self.genome = path
+	
+	def set_protein(self, path):
+		self.status = 'protein'
+		self.protein = path
+
+	def set_hmm(self, path):
+		if self.protein is None:
+			print("Warning! I don't have a protein yet, so this HMM will be useless to me until I do!")
+		self.status = 'protein and hmm'
+		self.hmm = path
+		
+	def set_crystal(self, path):
+		self.status = 'crystal'
+		self.crystal = path
+
+	#Runs prodigal, compares translation tables and stores faa files
+	def genome_to_protein(self):
+		if self.genome is None:
+			print(self.name, "wasn't a declared as a genome! I can't make this into a protein!")
+		else:		
+			protein_output = os.path.normpath(self.output + "/predicted_proteins/" + self.basename + '.faa')
+					
+			if self.do_compress:
+				compress_level = "1"
+			else:
+				compress_level = "0"
+			
+			mn = pyrodigal_manager(file = self.genome, aa_out = protein_output, compare_against = [4], do_compress = compress_level)
+			mn.run_for_fastaai()
+			
+			self.trans_table = str(mn.trans_table)
+			
+			for prot in mn.excluded_seqs:
+				self.err_log += "Protein " + prot + " was observed to have >100K amino acids ( " + str(mn.excluded_seqs[prot]) + " AA found ). It will not be included in predicted proteins for this genome;"
+			
+			self.prepared_proteins = mn.labeled_proteins
+			
+			del mn
+			
+			#If there are zipped files leftover and we didn't want them, clean them up.
+			if self.do_compress:
+				self.set_protein(str(protein_output)+".gz")
+				#Clean up unzipped version on reruns
+				if os.path.exists(str(protein_output)):
+					os.remove(str(protein_output))
+			else:
+				self.set_protein(str(protein_output))
+				#Clean up a zipped version on reruns
+				if os.path.exists(str(protein_output)+".gz"):
+					os.remove(str(protein_output)+".gz")
+
+			self.prot_pred_time = datetime.datetime.now()
+					
+	#run hmmsearch on a protein	
+	def protein_to_hmm(self):
+		if self.protein is None:
+			print(self.basename, "wasn't a declared as a protein! I can't make this into an HMM!")
+		else:
+
+			folder = os.path.normpath(self.output + "/hmms")
+						
+			hmm_output = os.path.normpath(folder +"/"+ self.basename + '.hmm')
+			
+			if self.prepared_proteins is None:
+				self.prepared_proteins, deflines = read_fasta(self.protein)
+
+			hmm_manager.run_for_fastaai(self.prepared_proteins, hmm_output)
+			
+			self.ran_hmmer = True
+			
+			if self.do_compress:
+				self.set_hmm(str(hmm_output)+".gz")
+				if os.path.exists(str(hmm_output)):
+					os.remove(str(hmm_output))
+			else:
+				self.set_hmm(str(hmm_output))
+				if os.path.exists(str(hmm_output)+".gz"):
+					os.remove(str(hmm_output)+".gz")
+					
+			self.hmm_search_time = datetime.datetime.now()
+	
+	#Translate tetramers to unique int32 indices.
+	def unique_kmer_simple_key(self, seq):
+		#num tetramers = len(seq) - 4 + 1, just make it -3.
+		n_kmers = len(seq) - 3
+		
+		#Converts the characters in a sequence into their ascii int value
+		as_ints = np.array([ord(i) for i in seq], dtype = np.int32)
+		
+		#create seq like 0,1,2,3; 1,2,3,4; 2,3,4,5... for each tetramer that needs a value
+		kmers = np.arange(4*n_kmers)
+		kmers = kmers % 4 + kmers // 4
+		
+		#Select the characters (as ints) corresponding to each tetramer all at once and reshape into rows of 4, 
+		#each row corresp. to a successive tetramer
+		kmers = as_ints[kmers].reshape((n_kmers, 4))
+		
+		#Given four 2-digit numbers, these multipliers work as offsets so that all digits are preserved in order when summed
+		mult = np.array([1000000, 10000, 100, 1], dtype = np.int32)
+		
+		#the fixed values effectively offset the successive chars of the tetramer by 2 positions each time; 
+		#practically, this is concatenation of numbers
+		#Matrix mult does this for all values at once.
+		return np.unique(np.dot(kmers, mult))
+
+	def load_hmm_and_filter_from_file(self):
+		prots = []
+		accs = []
+		scores = []
+		f = agnostic_reader(self.hmm)
+		for line in f:
+			if line.startswith("#"):
+				continue
+			else:
+				segs = line.strip().split()
+				
+				if len(segs) < 9:
+					continue
+				
+				prots.append(segs[0])
+				accs.append(segs[3])
+				scores.append(segs[8])
+			
+		f.close()
+		
+		if len(prots) < 1:
+			self.best_hits = {}
+		
+		hmm_file = np.transpose(np.array([prots, accs, scores]))
+		
+		#hmm_file = np.loadtxt(hmm_file_name, comments = '#', usecols = (0, 3, 8), dtype=(str))
+		#Sort the hmm file based on the score column in descending order.
+		hmm_file = hmm_file[hmm_file[:,2].astype(float).argsort()[::-1]]
+		
+		#Identify the first row where each gene name appears, after sorting by score; 
+		#in effect, return the highest scoring assignment per gene name
+		#Sort the indices of the result to match the score-sorted table instead of alphabetical order of gene names
+		hmm_file = hmm_file[np.sort(np.unique(hmm_file[:,0], return_index = True)[1])]
+		
+		#Filter the file again for the unique ACCESSION names, since we're only allowed one gene per accession, I guess?
+		#Don't sort the indices, we don't care about the scores anymore.
+		hmm_file = hmm_file[np.unique(hmm_file[:,1], return_index = True)[1]]
+		
+		sql_friendly_names = [i.replace(".", "_") for i in hmm_file[:,1]]
+		self.best_hits = dict(zip(hmm_file[:,0], sql_friendly_names))
+			
+	#This should consider the domain by majority vote...
+	def prot_and_hmm_to_besthits(self):
+		if self.ran_hmmer:
+			#Manager has a filter built in.
+			self.best_hits = hmm_manager.best_hits
+		else:
+			#Load the best hits file via old numpy method.
+			self.load_hmm_and_filter_from_file()
+		
+		hit_count = 0
+				
+		#from pyrodigal predictions or HMM intermediate production, the sequences are already in mem and don't need read in.
+		if self.prepared_proteins is None:
+			#But otherwise, we need to read them in.
+			self.prepared_proteins, deflines = read_fasta(self.protein)
+		
+		self.protein_kmer_count = {}
+		self.best_hits_kmers = {}
+		
+		if self.crystalize:
+			crystal_record = []
+			
+		#Kmerize proteins and record metadata
+		for protein in self.prepared_proteins:
+			if protein in self.best_hits:
+				accession = self.best_hits[protein]
+				
+				if self.crystalize:
+					crystal_record.append(str(protein)+"\t"+str(accession)+"\t"+str(self.prepared_proteins[protein])+"\n")
+				
+				kmer_set = self.unique_kmer_simple_key(self.prepared_proteins[protein])
+				self.protein_kmer_count[accession] = kmer_set.shape[0]
+				self.protein_count += 1
+				self.best_hits_kmers[accession] = kmer_set
+				hit_count += 1
+				
+			#Free the space either way
+			self.prepared_proteins[protein] = None
+			
+		if self.crystalize:
+			#only make a crystal if it actually has content.
+			if len(crystal_record) > 0:
+				crystal_path = os.path.normpath(self.output + "/crystals/" + self.basename + '_faai_crystal.txt')
+				crystal_record = "".join(crystal_record)
+				
+				if self.do_compress:
+					crystal_record = crystal_record.encode()
+					crystal_writer = gzip.open(crystal_path+".gz", "wb")
+					crystal_writer.write(crystal_record)
+					crystal_writer.close()
+				else:
+					crystal_writer = open(crystal_path, "w")
+					crystal_writer.write(crystal_record)
+					crystal_writer.close()
+			
+		#Final free.
+		self.prepared_proteins = None
+			
+		#No HMM hits.
+		if hit_count == 0:
+			self.is_empty = True
+			
+		self.besthits_time = datetime.datetime.now()
+		self.status = "best hits found"
+
+	def preprocess(self):
+		self.init_time = datetime.datetime.now()
+		#default to 0 time.
+		self.prot_pred_time = self.init_time
+		self.hmm_search_time = self.init_time
+		self.besthits_time = self.init_time
+	
+		#There's no advancement stage for protein and HMM
+		if self.status == 'genome':
+			start_time = self.curtime()
+			#report = True
+			if self.start_time is None:
+				self.start_time = start_time
+			
+			if self.initial_state == "protein+HMM":
+				self.initial_state = "genome"
+			
+			self.genome_to_protein()
+			
+		if self.status == 'protein':
+			start_time = self.curtime()
+			#report = True
+			if self.start_time is None:
+				self.start_time = start_time
+				
+			if self.initial_state == "protein+HMM":
+				self.initial_state = "protein"
+			
+			self.protein_to_hmm()
+			
+		if self.status == 'protein and hmm':
+			start_time = self.curtime()
+			
+			if self.start_time is None:
+				self.start_time = start_time
+				
+			self.prot_and_hmm_to_besthits()
+		
+		#Add an end time if either genome -> protein -> HMM or protein -> HMM happened.
+		if self.start_time is not None:
+			end_time = self.curtime()
+			self.end_time = end_time
+		else:
+			#Start was protein+HMM. There was no runtime, and intitial state is p+hmm
+			#self.initial_state = "protein+HMM"
+			self.start_time = "N/A"
+			self.end_time = "N/A"
+			
+		#Protein not generated on this run.
+		if self.trans_table is None:
+			self.trans_table = "unknown"
+		
+		self.partial_timings()
+
+'''
+Utility functions
+'''
+def prepare_directories(output, status, build_or_query, make_crystals = False):
+	preparation_successful = True
+	
+	if not os.path.exists(output):
+		try:
+			os.mkdir(output)
+		except:
+			print("")
+			print("FastAAI tried to make output directory: '"+ output + "' but failed.")
+			print("")
+			print("Troubleshooting:")
+			print("")
+			print("    (1) Do you have permission to create directories in the location you specified?")
+			print("    (2) Did you make sure that all directories other than", os.path.basename(output), "already exist?")
+			print("")
+			preparation_successful = False
+	
+	if preparation_successful:
+		try:
+			if status == 'genome':
+				if not os.path.exists(os.path.normpath(output + "/" + "predicted_proteins")):
+					os.mkdir(os.path.normpath(output + "/" + "predicted_proteins"))
+				if not os.path.exists(os.path.normpath(output + "/" + "hmms")):
+					os.mkdir(os.path.normpath(output + "/" + "hmms"))
+			
+			if status == 'protein':
+				if not os.path.exists(os.path.normpath(output + "/" + "hmms")):
+					os.mkdir(os.path.normpath(output + "/" + "hmms"))
+					
+			if make_crystals:
+				if not os.path.exists(os.path.normpath(output + "/" + "crystals")):
+					os.mkdir(os.path.normpath(output + "/" + "crystals"))
+			
+			if build_or_query == "build":
+				if not os.path.exists(os.path.normpath(output + "/" + "database")):
+					os.mkdir(os.path.normpath(output + "/" + "database"))
+			
+			if build_or_query == "query":		
+				if not os.path.exists(os.path.normpath(output + "/" + "results")):
+					os.mkdir(os.path.normpath(output + "/" + "results"))
+
+			
+		except:
+			print("FastAAI was able to create or find", output, "but couldn't make directories there.")
+			print("")
+			print("This shouldn't happen. Do you have permission to write to that directory?")
+			
+		
+	return preparation_successful	
+
+def find_hmm():
+	hmm_path = None
+	try:
+		#Try to locate the data bundled as it would be with a pip/conda install.
+		script_path = os.path.dirname(sys.modules['fastAAI_HMM_models'].__file__)
+		if len(script_path) == 0:
+			script_path = "."
+		hmm_complete_model = os.path.abspath(os.path.normpath(script_path + '/00.Libraries/01.SCG_HMMs/Complete_SCG_DB.hmm'))
+		hmm_path = str(hmm_complete_model)
+		#Check that the file exists or fail to the except.
+		fh = open(hmm_path)
+		fh.close()
+	except:
+		#Look in the same dir as the script; old method/MiGA friendly
+		script_path = os.path.dirname(__file__)
+		if len(script_path) == 0:
+			script_path = "."
+		hmm_complete_model = os.path.abspath(os.path.normpath(script_path +"/"+ "00.Libraries/01.SCG_HMMs/Complete_SCG_DB.hmm"))
+		hmm_path = str(hmm_complete_model)
+	
+	return hmm_path
+
+#Build DB from genomes
+	
+def unique_kmers(seq, ksize):
+	n_kmers = len(seq) - ksize + 1
+	kmers = []
+	for i in range(n_kmers):
+		kmers.append(kmer_index[seq[i:i + ksize]])
+	#We care about the type because we're working with bytes later.
+	return np.unique(kmers).astype(np.int32)
+
+def split_seq(seq, num_grps):
+	newseq = []
+	splitsize = 1.0/num_grps*len(seq)
+	for i in range(num_grps):
+		newseq.append(seq[int(round(i*splitsize)):int(round((i+1)*splitsize))])
+	return newseq
+	
+#gives the max and min index needed to split a list of (max_val) genomes into 
+def split_indicies(max_val, num_grps):
+	newseq = []
+	splitsize = 1.0/num_grps*max_val
+	for i in range(num_grps):
+		newseq.append(((round(i*splitsize)), round((i+1)*splitsize)))
+	return newseq
+	
+def split_seq_indices(seq, num_grps):
+	newseq = []
+	splitsize = 1.0/num_grps*len(seq)
+	for i in range(num_grps):
+		newseq.append((int(round(i*splitsize)), int(round((i+1)*splitsize)),))
+	return newseq
+	
+	
+def list_to_index_dict(list):
+	result = {}
+	counter = 0
+	for item in list:
+		result[item] = counter
+		counter += 1
+	return result
+	
+	
+def rev_list_to_index_dict(list):
+	result = {}
+	counter = 0
+	for item in list:
+		result[counter] = item
+		counter += 1
+	return result
+	
+def generate_accessions_index(forward = True):
+	acc_list = ['PF01780_19', 'PF03948_14', 'PF17144_4', 'PF00830_19', 'PF00347_23', 'PF16906_5', 'PF13393_6',
+		'PF02565_15', 'PF01991_18', 'PF01984_20', 'PF00861_22', 'PF13656_6', 'PF00368_18', 'PF01142_18', 'PF00312_22', 'PF02367_17',
+		'PF01951_16', 'PF00749_21', 'PF01655_18', 'PF00318_20', 'PF01813_17', 'PF01649_18', 'PF01025_19', 'PF00380_19', 'PF01282_19',
+		'PF01864_17', 'PF01783_23', 'PF01808_18', 'PF01982_16', 'PF01715_17', 'PF00213_18', 'PF00119_20', 'PF00573_22', 'PF01981_16',
+		'PF00281_19', 'PF00584_20', 'PF00825_18', 'PF00406_22', 'PF00177_21', 'PF01192_22', 'PF05833_11', 'PF02699_15', 'PF01016_19',
+		'PF01765_19', 'PF00453_18', 'PF01193_24', 'PF05221_17', 'PF00231_19', 'PF00416_22', 'PF02033_18', 'PF01668_18', 'PF00886_19',
+		'PF00252_18', 'PF00572_18', 'PF00366_20', 'PF04104_14', 'PF04919_12', 'PF01912_18', 'PF00276_20', 'PF00203_21', 'PF00889_19',
+		'PF02996_17', 'PF00121_18', 'PF01990_17', 'PF00344_20', 'PF00297_22', 'PF01196_19', 'PF01194_17', 'PF01725_16', 'PF00750_19',
+		'PF00338_22', 'PF00238_19', 'PF01200_18', 'PF00162_19', 'PF00181_23', 'PF01866_17', 'PF00709_21', 'PF02006_16', 'PF00164_25',
+		'PF00237_19', 'PF01139_17', 'PF01351_18', 'PF04010_13', 'PF06093_13', 'PF00828_19', 'PF02410_15', 'PF01176_19', 'PF02130_17',
+		'PF01948_18', 'PF01195_19', 'PF01746_21', 'PF01667_17', 'PF03874_16', 'PF01090_19', 'PF01198_19', 'PF01250_17', 'PF17136_4',
+		'PF06026_14', 'PF03652_15', 'PF04019_12', 'PF01201_22', 'PF00832_20', 'PF01264_21', 'PF03840_14', 'PF00831_23', 'PF00189_20',
+		'PF02601_15', 'PF01496_19', 'PF00411_19', 'PF00334_19', 'PF00687_21', 'PF01157_18', 'PF01245_20', 'PF01994_16', 'PF01632_19',
+		'PF00827_17', 'PF01015_18', 'PF00829_21', 'PF00410_19', 'PF00833_18', 'PF00935_19', 'PF01992_16']
+	if forward:
+		list_of_poss_accs = list_to_index_dict(acc_list)
+	else:
+		list_of_poss_accs = rev_list_to_index_dict(acc_list)
+		
+	return list_of_poss_accs
+
+#Build or add to a FastAAI DB
+def build_db_opts():
+	parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter,
+			description='''
+	This FastAAI module allows you to create a FastAAI database from one or many genomes, proteins, or proteins and HMMs, or add these files to an existing one.
+	
+	Supply genomes OR proteins OR proteins AND HMMs as inputs.
+	
+	If you supply genomes, FastAAI will predict proteins from them, and HMMs will be created from those proteins
+	If you supply only proteins, FastAAI will create HMM files from them, searching against FastAAI's internal database
+	If you supply proteins AND HMMs, FastAAI will directly use them to build the database.\n
+	You cannot supply both genomes and proteins
+	''')
+
+	parser.add_argument('-g', '--genomes',  dest = 'genomes', default = None, help =  'A directory containing genomes in FASTA format.')
+	parser.add_argument('-p', '--proteins', dest = 'proteins', default = None, help = 'A directory containing protein amino acids in FASTA format.')
+	parser.add_argument('-m', '--hmms',     dest = 'hmms', default = None, help =     'A directory containing the results of an HMM search on a set of proteins.')
+	parser.add_argument('-d', '--database', dest = 'db_name', default = "FastAAI_database.sqlite.db", help =  'The name of the database you wish to create or add to. The database will be created if it doesn\'t already exist and placed in the output directory. FastAAI_database.sqlite.db by default.')
+	
+	parser.add_argument('-o', '--output',   dest = 'output', default = "FastAAI", help = 'The directory to place the database and any protein or HMM files FastAAI creates. By default, a directory named "FastAAI" will be created in the current working directory and results will be placed there.')
+						
+	parser.add_argument('--threads',  dest = 'threads', type=int, default = 1, help = 'The number of processors to use. Default 1.')
+	parser.add_argument('--verbose',        dest = 'verbose', action='store_true', help = 'Print minor updates to console. Major updates are printed regardless.')
+	parser.add_argument('--compress', dest = "do_comp", action = 'store_true', help = 'Gzip compress generated proteins, HMMs. Off by default.')
+
+	args, unknown = parser.parse_known_args()
+	
+	return parser, args
+	
+def run_build(input_file):
+	input_file.preprocess()
+	if len(input_file.best_hits_kmers) < 1:
+		input_file.best_hits_kmers = None
+		input_file.err_log += " This file did not successfully complete. No SCPs could be found."
+		
+	return input_file 
+	
+def acc_transformer_init(db, tempdir_path):
+	sqlite3.register_converter("array", convert_array)
+	global indb
+	indb = db
+	global temp_dir
+	temp_dir = tempdir_path
+	global ok
+	ok = generate_accessions_index()
+	
+def acc_transformer(acc_name):
+	source = sqlite3.connect(indb)
+	scurs = source.cursor()
+	
+	data = scurs.execute("SELECT * FROM {acc}_genomes".format(acc=acc_name)).fetchall()
+	
+	scurs.close()
+	source.close()
+	
+	reformat = {}
+	
+	for row in data:
+		genome, kmers = row[0], np.frombuffer(row[1], dtype=np.int32)
+		for k in kmers:
+			if k not in reformat:
+				reformat[k] = []
+			reformat[k].append(genome)
+	
+	data = None
+	
+	to_add = []
+	for k in reformat:
+		as_bytes = np.array(reformat[k], dtype = np.int32)
+		as_bytes = as_bytes.tobytes()
+		reformat[k] = None
+		to_add.append((int(k), as_bytes,))
+	
+	my_acc_db = os.path.normpath(temp_dir + "/"+acc_name+".db")
+	
+	if os.path.exists(my_acc_db):
+		os.remove(my_acc_db)
+	
+	my_db  = sqlite3.connect(my_acc_db)
+	curs = my_db.cursor()
+	curs.execute("CREATE TABLE {acc} (kmer INTEGER PRIMARY KEY, genomes array)".format(acc=acc_name))
+	my_db.commit()
+	
+	curs.executemany("INSERT INTO {acc} VALUES (?, ?)".format(acc = acc_name), to_add)
+	
+	my_db.commit()
+	
+	to_add = None
+	
+	curs.execute("CREATE INDEX {acc}_index ON {acc} (kmer)".format(acc=acc_name))
+	my_db.commit()
+	
+	curs.close()
+	my_db.close()
+	
+	return [my_acc_db, acc_name]
+	
+def build_db(genomes, proteins, hmms, db_name, output, threads, verbose, do_compress):
+	success = True
+	
+	imported_files = fastaai_file_importer(genomes = genomes, proteins = proteins, hmms = hmms, output = output, compress = do_compress)
+	imported_files.determine_inputs()
+	
+	if imported_files.error:
+		print("Exiting FastAAI due to input file error.")
+		quit()
+
+	good_to_go = prepare_directories(output, imported_files.status, "query")
+	
+	db_path = os.path.normpath(output + "/database")
+	if not os.path.exists(db_path):
+		os.mkdir(db_path)
+	
+	if not good_to_go:
+		print("Exiting FastAAI")
+		sys.exit()
+	
+	print("")
+
+	hmm_path = find_hmm()
+	
+	#Check if the db contains path info. Incl. windows version.
+	if "/" not in db_name and "\\" not in db_name:
+		final_database = os.path.normpath(output + "/database/" + db_name)
+	else:
+		#If the person insists that the db has a path, let them.
+		final_database = db_name
+		
+	#We'll skip trying this if the file already exists.
+	existing_genome_IDs = None
+	try:
+		if os.path.exists(final_database):
+			parent = sqlite3.connect(final_database)
+			curs = parent.cursor()
+			
+			existing_genome_IDs = {}
+			sql_command = "SELECT genome, gen_id FROM genome_index"
+			for result in curs.execute(sql_command).fetchall():
+				genome = result[0]
+				id = int(result[1])
+				existing_genome_IDs[genome] = id
+			
+			curs.close()
+			parent.close()
+	except:
+		print("You specified an existing file to be a database, but it does not appear to be a FastAAI database.")
+		print("FastAAI will not be able to continue. Please give FastAAI a different database name and continue.")
+		print("Exiting.")
+		success = False
+	
+	if success:
+		hmm_file = find_hmm()
+		if existing_genome_IDs is not None:
+			genome_idx = max(list(existing_genome_IDs.values()))+1
+		else:
+			existing_genome_IDs = {}
+			genome_idx = 0
+		
+		#return_to
+		td = tempfile.mkdtemp()
+		#if not os.path.exists(td):
+		#	os.mkdir(td)
+		
+		temp_db = os.path.normpath(td+"/FastAAI_temp_db.db")
+
+		if os.path.exists(temp_db):
+			os.remove(temp_db)
+		
+		sqlite3.register_converter("array", convert_array)
+		worker = sqlite3.connect(temp_db)
+		wcurs = worker.cursor()
+		wcurs.execute("CREATE TABLE genome_index (genome text, gen_id integer, protein_count integer)")
+		wcurs.execute("CREATE TABLE genome_acc_kmer_counts (genome integer, accession integer, count integer)")
+		ok = generate_accessions_index()
+		for t in ok:
+			wcurs.execute("CREATE TABLE " + t + "_genomes (genome INTEGER PRIMARY KEY, kmers array)")
+			
+		worker.commit()
+
+		new_gens = []
+		new_gak = []
+		accs_seen = {}
+		if verbose:
+			tracker = progress_tracker(total = len(imported_files.in_files), message = "Processing inputs")
+		else:
+			print("Processing inputs")
+			
+		#Only build_db makes a log.
+		if not os.path.exists(os.path.normpath(output + "/" + "logs")):
+			os.mkdir(os.path.normpath(output + "/" + "logs"))
+		
+		logger = open(os.path.normpath(output+"/logs/"+"FastAAI_preprocessing_log.txt"), "a")
+		print("file", "start_date", "end_date", "starting_format", 
+		"prot_prediction_time", "trans_table", "hmm_search_time", "besthits_time", 
+		"errors", sep = "\t", file = logger)
+		
+		pool = multiprocessing.Pool(threads, initializer = hmm_preproc_initializer, 
+		initargs = (hmm_file, do_compress,))
+		
+		for result in pool.imap(run_build, imported_files.in_files):
+			#log data, regardless of kind
+			print(result.basename, result.start_time, result.end_time, result.initial_state, 
+			result.prot_pred_time, result.trans_table, result.hmm_search_time, result.besthits_time,
+			result.err_log, sep = "\t", file = logger)
+			
+			if result.best_hits_kmers is not None:
+				genome_name = result.original_name
+				
+				if genome_name in existing_genome_IDs:
+					print(genome_name, "Already present in final database and will be skipped.")
+					print("")
+				else:
+					protein_count = result.protein_count
+					for acc_name in result.best_hits_kmers:
+						if acc_name not in accs_seen:
+							accs_seen[acc_name] = 0
+						acc_id = ok[acc_name]
+						kmer_ct = result.protein_kmer_count[acc_name]
+						kmers = result.best_hits_kmers[acc_name]
+						kmers = kmers.tobytes()
+						wcurs.execute("INSERT INTO {acc}_genomes VALUES (?, ?)".format(acc=acc_name), (genome_idx, kmers,))
+						new_gak.append((genome_idx, acc_id, kmer_ct,))
+						
+					new_gens.append((genome_name, genome_idx, protein_count,))
+					genome_idx += 1
+					
+					worker.commit()
+					
+			if verbose:	
+				tracker.update()
+	
+		pool.close()
+		
+		logger.close()
+		
+		wcurs.executemany("INSERT INTO genome_index VALUES (?,?,?)", new_gens)
+		wcurs.executemany("INSERT INTO genome_acc_kmer_counts VALUES (?,?,?)", new_gak)
+		worker.commit()
+		
+		wcurs.close()
+		worker.close()
+		
+		accs_seen = list(accs_seen.keys())
+		
+		parent = sqlite3.connect(final_database)
+		curs = parent.cursor()
+				
+		curs.execute("attach '" + temp_db + "' as worker")
+		#initialize if needed.
+		curs.execute("CREATE TABLE IF NOT EXISTS genome_index (genome text, gen_id integer, protein_count integer)")
+		curs.execute("CREATE TABLE IF NOT EXISTS genome_acc_kmer_counts (genome integer, accession integer, count integer)")
+		
+		curs.execute("INSERT INTO genome_index SELECT * FROM worker.genome_index")
+		curs.execute("INSERT INTO genome_acc_kmer_counts SELECT * FROM worker.genome_acc_kmer_counts")
+		curs.execute("CREATE INDEX IF NOT EXISTS kmer_acc ON genome_acc_kmer_counts (genome, accession);")
+		parent.commit()
+		
+		if verbose:
+			tracker = progress_tracker(total = len(accs_seen), message = "Collecting results")
+		else:
+			print("Collecting results")
+		
+		pool = multiprocessing.Pool(threads, initializer = acc_transformer_init, 
+		initargs = (temp_db, td,))
+		
+		for result in pool.imap_unordered(acc_transformer, accs_seen):
+			database, accession = result[0], result[1]
+			curs.execute("CREATE TABLE IF NOT EXISTS {acc} (kmer INTEGER PRIMARY KEY, genomes array)".format(acc=accession))
+			curs.execute("CREATE TABLE IF NOT EXISTS {acc}_genomes (genome INTEGER PRIMARY KEY, kmers array)".format(acc=accession))
+			curs.execute("CREATE INDEX IF NOT EXISTS {acc}_index ON {acc}(kmer)".format(acc=accession))
+			
+			#Get the genomes from worker db.
+			curs.execute("INSERT INTO {acc}_genomes SELECT * FROM worker.{acc}_genomes".format(acc=accession))
+			
+			parent.commit()
+			
+			accdb = sqlite3.connect(database)
+			acc_curs = accdb.cursor()
+			
+			to_update = acc_curs.execute("SELECT kmer, genomes, genomes FROM {acc}".format(acc=accession)).fetchall()
+			
+			acc_curs.close()
+			accdb.close()
+			
+			update_concat_sql = "INSERT INTO {acc} VALUES (?,?) ON CONFLICT(kmer) DO UPDATE SET genomes=genomes || (?)".format(acc=accession)
+			#ON CONFLICT(kmer) DO UPDATE SET genomes=genomes || acc.{acc}.genomes;".format(acc=accession)
+			#print(update_concat_sql)
+			curs.executemany(update_concat_sql, to_update)
+			
+			parent.commit()
+			
+			os.remove(database)
+			
+			if verbose:
+				tracker.update()
+		
+		pool.close()
+		
+		curs.execute("detach worker")
+		
+		parent.commit()
+		
+		curs.close()
+		parent.close()	
+		
+		os.remove(temp_db)
+	try:
+		if len(os.listdir(td)) == 0:
+			shutil.rmtree(td)
+	except:
+		pass
+
+	if success:
+		print("Database build complete!")
+		
+	return success
+
+def file_v_db_initializer(tgak, tgt_names, tgt_cts, hmm_file, do_compress, tgt_ct, sd, out, style, in_mem, build_q, tdb):
+	#num_tgts, self.do_sd, self.output, self.style, self.as_mem_db, self.do_db_build
+	global _tdb
+	_tdb = tdb
+	
+	global _tgt_gak
+	_tgt_gak = tgak
+	
+	global _tname
+	_tname = tgt_names
+	
+	global _tct
+	_tct = tgt_cts
+	
+	global hmm_manager
+	hmm_manager = pyhmmer_manager(do_compress)
+	hmm_manager.load_hmm_from_file(hmm_file)
+	
+	global num_tgts
+	num_tgts = tgt_ct
+	
+	global _do_sd
+	_do_sd = sd
+	
+	global out_style
+	out_style = style
+	
+	global out_base
+	out_base = out
+	
+	global db_is_in_mem
+	db_is_in_mem = in_mem
+	
+	global make_query_db
+	make_query_db = build_q
+	
+	return _tdb, _tgt_gak, _tname, _tct, hmm_manager, num_tgts, _do_sd, out_base, out_style, db_is_in_mem, make_query_db
+	
+def file_v_db_worker(query_args):
+	#query info for this particular query
+	in_file = query_args[0]
+	
+	in_file.preprocess()
+	
+	qname = in_file.basename
+	
+	do_sd = _do_sd
+	
+	#std dev. calcs are not meaningful with matrix style output.
+	if out_style == "matrix":
+		do_sd = False
+	
+	if do_sd:
+		results = []
+		shared_acc_counts = []
+	else:
+		results = np.zeros(shape = num_tgts, dtype = np.float_)
+		shared_acc_counts = np.zeros(shape = num_tgts, dtype = np.int32)
+	
+	if db_is_in_mem:
+		#The connection is already given as MDB if the db is in mem
+		tconn = _tdb
+	else:
+		#db is on disk and the connection has to be established.
+		tconn = sqlite3.connect(_tdb)
+		
+	tcurs = tconn.cursor()
+		
+	#This is a difference from the DB-first method.
+	acc_idx = generate_accessions_index(forward = True)
+	
+	genome_lists = {}
+	
+	tcurs.row_factory = lambda cursor, row: row[0]
+	
+	
+	if make_query_db:
+		ret = [qname, None, []]
+	else:
+		ret = [qname, None, None]
+	
+	#We need to purge accsessions not in tgt.
+	for acc in in_file.best_hits_kmers:
+		one = in_file.best_hits_kmers[acc]
+		acc_id = acc_idx[acc]
+		
+		if make_query_db:
+			ret[2].append((qname, acc_id, one.tobytes(),))
+		
+		#Check working.
+		if acc_id in _tgt_gak:
+
+			kmer_ct = one.shape[0]
+
+			if do_sd:
+				hits = np.zeros(shape = num_tgts, dtype = np.int32)
+				hits[np.nonzero(_tgt_gak[acc_id])] = 1
+				shared_acc_counts.append(hits)
+			else:
+				shared_acc_counts[np.nonzero(_tgt_gak[acc_id])] += 1
+			
+			#SQL has a max binding size of 999, for some reason.
+			if kmer_ct > 998:
+				#Each kmer needs to be a tuple.
+				these_kmers = [(int(kmer),) for kmer in one]
+			
+				temp_name = "_" + qname +"_" + acc
+				temp_name = temp_name.replace(".", "_")
+				
+				tcurs.execute("CREATE TEMP TABLE " + temp_name + " (kmer INTEGER)")
+				tconn.commit()
+				insert_table = "INSERT INTO " + temp_name + " VALUES (?)"
+				tcurs.executemany(insert_table, these_kmers)
+				tconn.commit()
+				join_and_select_sql = "SELECT genomes FROM " + temp_name + " INNER JOIN " + acc + " ON "+ temp_name+".kmer = " + acc+".kmer;"
+
+				set = tcurs.execute(join_and_select_sql).fetchall()
+			else:
+				#kmers must be a list, not a tuple.
+				these_kmers = [int(kmer) for kmer in one]
+				select = "SELECT genomes FROM " + acc + " WHERE kmer IN ({kmers})".format(kmers=','.join(['?']*len(these_kmers)))
+				
+				set = tcurs.execute(select, these_kmers).fetchall()
+				
+			#join results into one bytestring.
+			set = b''.join(set)
+			
+			these_intersections = np.bincount(np.frombuffer(set, dtype = np.int32), minlength = num_tgts)
+			set = None
+			#Add tgt kmer counts to query kmer counts, find union size based on intersection size, cald jacc
+			jacc = np.divide(these_intersections, np.subtract(np.add(_tgt_gak[acc_id], kmer_ct), these_intersections))
+			
+			if do_sd:
+				results.append(jacc)
+			else:
+				results += jacc
+			
+	tcurs.row_factory = None
+	tcurs.close()
+	
+	if do_sd:
+		results = np.vstack(results)
+		has_accs = np.vstack(shared_acc_counts)
+		
+		shared_acc_counts = np.sum(has_accs, axis = 0)
+		
+		#final jacc_means
+		jaccard_averages = np.divide(np.sum(results, axis = 0), shared_acc_counts)
+		
+		aai_ests = numpy_kaai_to_aai(jaccard_averages)
+		
+		#find diffs from means; this includes indicies corresponding to unshared SCPs that should not be included. 
+		results = results - jaccard_averages
+		
+		#fix those corresponding indicies to not contribute to the final SD.
+		results[np.nonzero(has_accs == 0)] = 0
+		
+		#Square them
+		results = np.square(results)
+		#Sum squares and divide by shared acc. count, the sqrt to get SD.
+		jaccard_SDs = np.sqrt(np.divide(np.sum(results, axis = 0), shared_acc_counts))
+		jaccard_SDs = np.round(jaccard_SDs, 4).astype(str)
+				
+	else:
+		#other condition.
+		jaccard_SDs = None
+		jaccard_averages = np.divide(results, shared_acc_counts)
+		#we don't want to pass char arrays to main, so skip this here and do it in main instead.
+		if out_style != "matrix":
+			aai_ests = numpy_kaai_to_aai(jaccard_averages)
+		
+	del results
+
+	#Since the outputs go to separate files, it makes more sense to do them within the worker processes instead of in main.
+	if out_style == "tsv":
+		no_hit = np.where(shared_acc_counts == 0)
+		
+		possible_hits = np.minimum(len(in_file.best_hits_kmers), _tct).astype(str)
+		jaccard_averages = np.round(jaccard_averages, 4).astype(str)
+		shared_acc_counts = shared_acc_counts.astype(str)
+		
+		jaccard_averages[no_hit] = "N/A"
+		aai_ests[no_hit] = "N/A"
+		shared_acc_counts[no_hit] = "N/A"
+		possible_hits[no_hit] = "N/A"
+			
+		output_name = os.path.normpath(out_base + "/"+qname+"_results.txt")
+	
+		out = open(output_name, "w")
+		out.write("query\ttarget\tavg_jacc_sim\tjacc_SD\tnum_shared_SCPs\tposs_shared_SCPs\tAAI_estimate\n")
+		if do_sd:
+			jaccard_SDs[no_hit] = "N/A"
+			for i in range(0, len(aai_ests)):
+				out.write(qname+"\t"+_tname[i]+"\t"+jaccard_averages[i]+"\t"+jaccard_SDs[i]+"\t"+shared_acc_counts[i]+"\t"+possible_hits[i]+"\t"+aai_ests[i]+"\n")
+		else:			
+			for i in range(0, len(aai_ests)):
+				out.write(qname+"\t"+_tname[i]+"\t"+jaccard_averages[i]+"\t"+"N/A"+"\t"+shared_acc_counts[i]+"\t"+possible_hits[i]+"\t"+aai_ests[i]+"\n")
+		out.close()
+
+		
+	#We're just gonna pass this back to the main to print.
+	if out_style == "matrix":
+		ret[1] = jaccard_averages
+	
+	return ret
+	
+#Handles both query and target types for a db vs db query
+class file_vs_db_query:
+	def __init__(self, in_memory = False, input_file_objects = None,
+	target = None, threads = 1, do_sd = False, output_base = "FastAAI", output_style = "tsv", 
+	build_db_from_queries = True, qdb_name = "Query_FastAAI_database.db", hmm_path = None, 
+	do_comp = True, verbose = True):
+		#files to work with
+		self.queries = input_file_objects
+		self.do_db_build = build_db_from_queries
+		self.dbname = qdb_name
+		
+		self.t = target
+		self.valids = None
+		
+		#Originally this was made to be a memory database only block of code, but just if/else one change makes it work on disk and it doesn't need a redev, then.
+		self.as_mem_db = in_memory
+		
+		self.t_conn = None
+		self.t_curs = None
+		
+		self.threads = threads
+		self.do_sd = do_sd
+
+		self.output_base = output_base
+		self.output = os.path.normpath(output_base + "/results")
+		self.style = output_style
+		
+		if hmm_path is not None:
+			self.hmm_path = hmm_path
+		else:
+			self.hmm_path = find_hmm()
+			
+		self.do_comp = do_comp
+		
+		self.verbose = verbose
+
+	'''
+	Workflow is:
+		load target db as mem (optional)
+		assess valid targets
+		create query db output (optional)
+		pass query args to workers
+		preproc query args
+		write results
+		fill query_db_out (optional)
+	'''
+		
+		
+	def open(self):
+		if self.as_mem_db:
+			self.t_conn = sqlite3.connect(':memory:')
+		else:
+			self.t_conn = sqlite3.connect(self.t)
+			
+		self.t_curs = self.t_conn.cursor()	
+		
+		if self.as_mem_db:
+			self.t_curs.execute("attach '" + self.t + "' as targets")
+			
+			self.t_curs.execute("CREATE TABLE genome_index AS SELECT * FROM targets.genome_index")
+			self.t_curs.execute("CREATE TABLE genome_acc_kmer_counts AS SELECT * FROM targets.genome_acc_kmer_counts")
+			self.t_curs.execute("CREATE INDEX t_gi ON genome_index (gen_id)")
+			self.t_curs.execute("CREATE INDEX t_gak ON genome_acc_kmer_counts (accession)")
+		
+		if self.as_mem_db:
+			table_sql = "SELECT name FROM targets.sqlite_master"
+		else:
+			table_sql = "SELECT name FROM sqlite_master"
+		
+		
+		ok = generate_accessions_index()
+		ok_names = set(list(ok.keys()))
+		successful_tables = []
+		
+		for name in self.t_curs.execute(table_sql).fetchall():
+			name = name[0]
+			if name in ok_names:
+				successful_tables.append(ok[name])
+				if self.as_mem_db:
+					self.t_curs.execute("CREATE TABLE " + name + " AS SELECT * FROM targets."+name)
+					self.t_curs.execute("CREATE INDEX "+name+"_index ON " + name+" (kmer)" )
+					
+		if self.as_mem_db:
+			self.t_conn.commit()
+			self.t_curs.execute("detach targets")
+			
+		self.valids = tuple(successful_tables)
+			
+	def close(self):
+		self.t_curs.close()
+		self.t_curs = None
+		
+	def clean_up(self):
+		self.t_conn.close()
+		self.t_conn = None
+		
+	def sqlite_table_schema(self, conn, name):
+		"""Return a string representing the table's CREATE"""
+		cursor = conn.execute("SELECT sql FROM sqlite_master WHERE name=?;", [name])
+		sql = cursor.fetchone()[0]
+		cursor.close()
+		return sql
+		
+	def execute(self):
+		print("FastAAI is running.")
+		tgt_id_res = self.t_curs.execute("SELECT * FROM genome_index ORDER BY gen_id").fetchall()
+		
+		tgt_ids = []
+		tgt_naming = []
+		tgt_counts = []
+		for r in tgt_id_res:
+			genome, id, prot_ct = r[0], r[1], r[2]
+			tgt_ids.append(genome)
+			tgt_naming.append(genome)
+			tgt_counts.append(prot_ct)
+			
+		num_tgts = len(tgt_ids)
+		tgt_counts = np.array(tgt_counts, dtype = np.int32)
+			
+		tgts_gak = {}
+		gak_sql = "SELECT * FROM genome_acc_kmer_counts WHERE accession in ({accs})".format(accs=','.join(['?']*len(self.valids)))
+		
+		for result in self.t_curs.execute(gak_sql, self.valids).fetchall():
+			genome, acc, ct = result[0], result[1], result[2]
+			if acc not in tgts_gak:
+				tgts_gak[acc] = np.zeros(num_tgts, dtype = np.int32)
+			tgts_gak[acc][genome] += ct
+			
+		#If the DB is a memory DB, we need to maintain the connection, but neither needs to maintain the curor in main.
+		self.close()
+			
+		query_groups = []
+		
+		for query_input in self.queries:
+			query_groups.append((query_input,))
+		
+		#And if it's a physical database, we do want to close it.
+		if not self.as_mem_db:
+			self.t_conn.close()
+		
+		num_queries = len(query_groups)
+		
+		if self.do_db_build:
+			sqlite3.register_converter("array", convert_array)
+			qdb_path = os.path.normpath(self.output_base + "/database/"+self.dbname)
+			if not os.path.exists(os.path.normpath(self.output_base + "/database")):
+				try:
+					os.mkdir(os.path.normpath(self.output_base + "/database"))
+				except:
+					print("Couldn't make database at", qdb_path)
+					self.do_db_build = False
+					
+			if os.path.exists(qdb_path):
+				print("Database for queries already exists. I can't make one at:", qdb_path)
+				self.do_db_build = False
+			else:
+				query_db_conn = sqlite3.connect(qdb_path)
+				q_curs = query_db_conn.cursor()
+				q_curs.execute("CREATE TABLE storage (genome INTEGER, accession INTEGER, kmers array)")
+				q_curs.execute("CREATE INDEX store_idx ON storage (genome, accession)")
+				query_genome_index = []
+				qgi_ct = 0
+				qg_gak = []
+		
+		if self.verbose:
+			tracker = progress_tracker(total = num_queries, message = "Calculating AAI...", one_line = True)
+		
+		if self.style == "matrix":
+			output_name = os.path.normpath(self.output + "/FastAAI_matrix.txt")
+			output = open(output_name, "w")
+			#needs target names.
+			print("query_genome", *tgt_ids, sep = "\t", file = output)
+		
+		#Need to pass these
+		
+		#both initializers will share this.
+		shared_args = [tgts_gak, tgt_naming, tgt_counts, self.hmm_path, self.do_comp, num_tgts, self.do_sd, self.output, 
+			self.style, self.as_mem_db, self.do_db_build]
+		
+		if self.as_mem_db:
+			shared_args.append(self.t_conn)
+			shared_args = tuple(shared_args)
+			pool = multiprocessing.Pool(self.threads, initializer = file_v_db_initializer, 
+			initargs = shared_args)
+		else:
+			#db is on disk,
+			shared_args.append(self.t)
+			shared_args = tuple(shared_args)
+			pool = multiprocessing.Pool(self.threads, initializer = file_v_db_initializer, 
+			initargs = shared_args)
+
+		for result in pool.imap(file_v_db_worker, query_groups):
+			if self.verbose:
+				tracker.update()
+			qname = result[0]
+			if self.style == "matrix":
+				printout = numpy_kaai_to_aai(result[1])
+				print(qname, *printout, sep = "\t", file = output)
+				
+			if self.do_db_build:
+				query_genome_index.append((qname, qgi_ct, len(result[2]),))
+				for row in result[2]:
+					num_kmers = int(len(row[2])/4)
+					qg_gak.append((qgi_ct, row[1], num_kmers,))
+				qgi_ct += 1
+				q_curs.executemany("INSERT INTO storage VALUES (?, ?, ?)", result[2])
+				query_db_conn.commit()
+				
+		pool.close()
+		
+		if self.style == "matrix":
+			output.close()
+			
+		if self.do_db_build:
+			q_curs.execute("CREATE TABLE genome_index (genome text, gen_id integer, protein_count integer)")
+			q_curs.execute("CREATE TABLE genome_acc_kmer_counts (genome integer, accession integer, count integer)")
+			q_curs.executemany("INSERT INTO genome_index VALUES (?,?,?)", query_genome_index)
+			q_curs.executemany("INSERT INTO genome_acc_kmer_counts VALUES (?,?,?)", qg_gak)
+			query_db_conn.commit()
+			
+			acc_id_to_name = generate_accessions_index(forward = False)
+			qgi_dict = {}
+			for tup in query_genome_index:
+				qgi_dict[tup[0]] = tup[1]
+			
+			accs_in_db = q_curs.execute("SELECT DISTINCT(accession) FROM genome_acc_kmer_counts").fetchall()
+			if self.verbose:
+				tracker = progress_tracker(total = len(accs_in_db), message = "Crafting database from query outputs.", one_line = True)
+			
+			for acc in accs_in_db:
+				acc = acc[0]
+				acc_name = acc_id_to_name[acc]
+				q_curs.execute("CREATE TABLE " + acc_name + " (kmer INTEGER PRIMARY KEY, genomes array)")
+				q_curs.execute("CREATE TABLE " + acc_name + "_genomes (genome INTEGER PRIMARY KEY, kmers array)")
+				data = q_curs.execute("SELECT genome, kmers FROM storage WHERE accession = ?", (acc,)).fetchall()
+				
+				ins = []
+				#group by kmer
+				kmers_by_gen = {}
+				for row in data:
+					gen = row[0]
+					gen = qgi_dict[gen]
+					kmers = np.frombuffer(row[1], dtype = np.int32)
+					ins.append((gen, kmers,))
+					for k in kmers:
+						#typecast
+						k = int(k)
+						if k not in kmers_by_gen:
+							kmers_by_gen[k] = []
+						kmers_by_gen[k].append(gen)
+						
+				data = None
+				
+				q_curs.executemany("INSERT INTO "+ acc_name + "_genomes VALUES (?,?)", ins)
+				
+				ins = []
+				for k in kmers_by_gen:
+					dat = kmers_by_gen[k]
+					dat = np.sort(np.array(dat, dtype = np.int32))
+					ins.append((k, dat.tobytes()))
+				
+				q_curs.executemany("INSERT INTO "+ acc_name + " VALUES (?,?)", ins)
+				
+				ins = None
+				
+				query_db_conn.commit()
+				
+				q_curs.execute("CREATE INDEX IF NOT EXISTS " + acc_name + "_index ON " + acc_name + " (kmer)")
+				
+				if self.verbose:
+					tracker.update()
+				
+				
+			q_curs.execute("CREATE INDEX IF NOT EXISTS kmer_acc ON genome_acc_kmer_counts (genome, accession);")
+			q_curs.execute("DROP INDEX store_idx")
+			q_curs.execute("DROP TABLE storage")
+			query_db_conn.commit()
+			q_curs.execute("VACUUM")
+			query_db_conn.commit()
+			q_curs.close()
+			query_db_conn.close()
+
+	#Actually run the thing.
+	def run(self):
+		self.open()
+		self.execute()
+		#Clean up the db connections; free the mem.
+		self.clean_up()
+	
+def numpy_kaai_to_aai(kaai_array):
+	#aai_hat = (-0.3087057 + 1.810741 * (np.exp(-(-0.2607023 * np.log(kaai))**(1/3.435))))*100
+	
+	#Protect the original jaccard averages memory item
+	aai_hat_array = kaai_array.copy()
+	
+	non_zero = np.where(aai_hat_array > 0)
+	is_zero  = np.where(aai_hat_array <= 0)
+	
+	#I broke this down into its original components
+	#Avoid zeroes in log - still actually works, but it produces warnings I don't want to see.
+	aai_hat_array[non_zero] = np.log(aai_hat_array[non_zero])
+	
+	aai_hat_array = np.multiply(np.subtract(np.multiply(np.exp(np.negative(np.power(np.multiply(aai_hat_array, -0.2607023), (1/3.435)))), 1.810741), 0.3087057), 100)
+	'''
+	Same as the above, broken down into easier-to-follow steps.
+	aai_hat_array = np.multiply(aai_hat_array, -0.2607023)
+	aai_hat_array = np.power(aai_hat_array, (1/3.435))
+	aai_hat_array = np.negative(aai_hat_array)
+	aai_hat_array = np.exp(aai_hat_array)
+	aai_hat_array = np.multiply(aai_hat_array, 1.810741)
+	aai_hat_array = np.subtract(aai_hat_array, 0.3087057)
+	aai_hat_array = np.multiply(aai_hat_array, 100)
+	'''
+	
+	#<30 and >90 values
+	smol = np.where(aai_hat_array < 30)
+	big  = np.where(aai_hat_array > 90)
+	
+	aai_hat_array = np.round(aai_hat_array, 2)
+	
+	#Convert to final printables
+	aai_hat_array = aai_hat_array.astype(str)
+	aai_hat_array[smol] = "<30%"
+	aai_hat_array[big]  = ">90%"
+	#The math of the above ends up with zero values being big, so we fix those.
+	aai_hat_array[is_zero] = "<30%"
+	
+	return aai_hat_array
+	
+#Also includes a multiply by 100 and type conversion compared to original - this is some silliness for saving memory.
+def numpy_kaai_to_aai_just_nums(kaai_array, as_float = False):
+	#aai_hat = (-0.3087057 + 1.810741 * (np.exp(-(-0.2607023 * np.log(kaai))**(1/3.435))))*100
+	
+	#Protect the original jaccard averages memory item
+	aai_hat_array = kaai_array.copy()
+	
+	non_zero = np.where(aai_hat_array > 0)
+	is_zero  = np.where(aai_hat_array <= 0)
+	
+	#I broke this down into its original components
+	#Avoid zeroes in log - still actually works, but it produces warnings I don't want to see.
+	aai_hat_array[non_zero] = np.log(aai_hat_array[non_zero])
+	
+	aai_hat_array = np.multiply(np.subtract(np.multiply(np.exp(np.negative(np.power(np.multiply(aai_hat_array, -0.2607023), (1/3.435)))), 1.810741), 0.3087057), 100)
+	'''
+	Same as the above, broken down into easier-to-follow steps.
+	aai_hat_array = np.multiply(aai_hat_array, -0.2607023)
+	aai_hat_array = np.power(aai_hat_array, (1/3.435))
+	aai_hat_array = np.negative(aai_hat_array)
+	aai_hat_array = np.exp(aai_hat_array)
+	aai_hat_array = np.multiply(aai_hat_array, 1.810741)
+	aai_hat_array = np.subtract(aai_hat_array, 0.3087057)
+	aai_hat_array = np.multiply(aai_hat_array, 100)
+	'''
+	
+	aai_hat_array = np.round(aai_hat_array, 2)
+	
+	#<30 and >90 values
+	smol = np.where(aai_hat_array < 30)
+	big  = np.where(aai_hat_array > 90)
+	
+	#We can find these later.
+	aai_hat_array[smol] = 15
+	aai_hat_array[big]  = 95
+	
+	if as_float:
+		aai_hat_array = np.round(aai_hat_array, 2)
+	else:
+		aai_hat_array = np.multiply(aai_hat_array, 100)
+		aai_hat_array = np.round(aai_hat_array, 2)
+		aai_hat_array = aai_hat_array.astype(np.int16)
+	
+	return aai_hat_array
+
+	
+def curtime():
+	time_format = "%d/%m/%Y %H:%M:%S"
+	timer = datetime.datetime.now()
+	time = timer.strftime(time_format)
+	return time
+
+#Perform a minimal-memory query of a target database from input files. Lighter weight function for low memory
+def sql_query_opts():
+	parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter,
+			description='''
+	This FastAAI module takes one or many genomes, proteins, or proteins and HMMs as a QUERY and searches them against an existing FastAAI database TARGET using SQL
+	If you only have a few genomes - or not enough RAM to hold the entire target database in memory - this is the probably the best option for you.
+	
+	To provide files, supply either a directory containing only one type of file (e.g. only genomes in FASTA format), a file containing paths to files of a type, 1 per line,
+	or a comma-separated list of files of a single type (no spaces)
+	
+	If you provide FastAAI with genomes or only proteins (not proteins and HMMs), this FastAAI module will produce the required protein and HMM files as needed
+	and place them in the output directory, just like it does while building a database. 
+	
+	Once these inputs are ready to be queried against the database (each has both a protein and HMM file), they will be processed independently, 1 per thread at a time.
+	
+	Note: Protein and HMM files generated during this query can be supplied to build a FastAAI database from proteins and HMMs using the build_db module, without redoing preprocessing.
+	''')
+			
+	parser.add_argument('-g', '--genomes',  dest = 'genomes', default = None,  help = 'Genomes in FASTA format.')
+	parser.add_argument('-p', '--proteins', dest = 'proteins', default = None, help = 'Protein amino acids in FASTA format.')
+	parser.add_argument('-m', '--hmms',     dest = 'hmms', default = None,     help = 'HMM search files produced by FastAAI on a set of proteins.')
+	
+	parser.add_argument('--target',   dest = 'target', default = None, help =   'A path to the FastAAI database you wish to use as the target')
+	
+	parser.add_argument('-o', '--output',   dest = 'output', default = "FastAAI", help = 'The directory where FastAAI will place the result of this query and any protein or HMM files it has to generate. By default, a directory named "FastAAI" will be created in the current working directory and results will be placed there.')
+	parser.add_argument('--output_style', dest = "style", default = 'tsv', help = "Either 'tsv' or 'matrix'. Matrix produces a simplified output of only AAI estimates.")
+	parser.add_argument('--do_stdev',   dest = "do_stdev", action='store_true',   help = 'Off by default. Calculate std. deviations on Jaccard indicies. Increases memory usage and runtime slightly. Does NOT change estimated AAI values at all.')
+	
+	parser.add_argument('--threads',  dest = 'threads', type=int, default = 1, help = 'The number of processors to use. Default 1.')
+	parser.add_argument('--verbose',        dest = 'verbose', action='store_true', help = 'Print minor updates to console. Major updates are printed regardless.')
+	
+	parser.add_argument('--in_memory', dest = "in_mem", action = 'store_true', help = 'Load the target database into memory before querying. Consumes more RAM, but is faster and reduces file I/O substantially.')
+	
+	parser.add_argument('--create_query_db', dest = "make_db", action = 'store_true', help = 'Create a query database from the genomes.')
+	parser.add_argument('--query_db_name', dest = "qdb_name", default = "Query_FastAAI_db.db", help = 'Name the query database. This file must not already exist.')
+	
+	parser.add_argument('--compress', dest = "do_comp", action = 'store_true', help = 'Gzip compress generated proteins, HMMs. Off by default.')
+
+	args, unknown = parser.parse_known_args()
+	
+	return parser, args
+
+def sql_query_thread_starter(kmer_cts, protein_cts):
+	global target_kmer_cts 
+	global target_protein_counts
+	target_kmer_cts = kmer_cts
+	target_protein_counts = protein_cts
+
+#took a function from fastaai 2.0
+class fastaai_file_importer:
+	def __init__(self, genomes = None, proteins = None, hmms = None, crystals = None, 
+	output = "FastAAI", compress = False, crystalize = False):
+		#genomes, prots, hmms can be supplied as either directory, a file with paths 1/line, or comma-sep paths. Type is determined automatically.
+		self.genomes = genomes
+		self.proteins = proteins
+		self.hmms = hmms
+		self.crystals = crystals
+		
+		self.genome_list = None
+		self.protein_list = None
+		self.hmm_list = None
+		self.crystal_list = None
+		
+		self.crystalize = crystalize
+		
+		#file base names.
+		self.identifiers = None
+		
+		self.error = False
+		
+		self.in_files = None
+		
+		self.status = "genome"
+		self.output = output
+		
+		self.do_comp = compress
+	
+	def retrieve_files(self, arg):	
+		done = False
+		files = []
+		names = []
+		#Case where a directory is supplied.
+		if os.path.isdir(arg):
+			for file in sorted(os.listdir(arg)):
+				#Retrieve file name
+				if file.endswith(".gz"):
+					name = os.path.splitext(os.path.basename(file[:-3]))[0]
+				else:
+					name = os.path.splitext(os.path.basename(file))[0]
+				
+				names.append(name)				
+				files.append(os.path.abspath(os.path.normpath(arg + '/' +file)))
+				
+			done = True
+				
+			
+		#Case where a file containing paths is supplied.
+		if os.path.isfile(arg):
+			handle = agnostic_reader(arg)
+			for line in handle:
+				file = line.strip()
+				if os.path.exists(file):
+					if file.endswith(".gz"):
+						name = os.path.splitext(os.path.basename(file[:-3]))[0]
+					else:
+						name = os.path.splitext(os.path.basename(file))[0]
+					
+					names.append(name)				
+					files.append(os.path.abspath(os.path.normpath(file)))
+					
+			handle.close()
+			done = True
+			
+			if len(names) == 0 and len(files) == 0:
+				#Try interpreting the file as a singular path.
+				done = False
+							
+		#Last check.
+		if not done:
+			for file in arg.split(","):
+				if os.path.exists(file):
+					if file.endswith(".gz"):
+						name = os.path.splitext(os.path.basename(file[:-3]))[0]
+					else:
+						name = os.path.splitext(os.path.basename(file))[0]
+					
+					names.append(name)				
+					files.append(os.path.abspath(os.path.normpath(file)))
+				
+		return files, names
+	
+	#Check if g/p/h
+	def determine_inputs(self):
+		if self.genomes is not None:
+			self.genome_list, self.identifiers = self.retrieve_files(self.genomes)
+			if self.proteins is not None or self.hmms is not None:
+				print("You can supply genomes or proteins or proteins and HMMS, but not genomes and anything else.")
+				self.error = True
+			
+		#Proteins, but no HMMs
+		if self.proteins is not None and self.hmms is None:
+			self.protein_list, self.identifiers = self.retrieve_files(self.proteins)
+			
+		if self.proteins is not None and self.hmms is not None:
+			self.protein_list, prot_names = self.retrieve_files(self.proteins)
+			self.hmm_list, hmm_names = self.retrieve_files(self.hmms)
+			
+			if len(self.protein_list) != len(self.hmm_list):
+				print("Different number of proteins and HMMs supplied. You must supply the same number of each, and they must be matched pairs.")
+				self.error = True
+			else:
+				all_same = True
+				for p, h in zip(prot_names, hmm_names):
+					if p != h:
+						all_same = False
+				
+				if all_same:
+					self.identifiers = prot_names
+					prot_names = None
+					hmm_names = None
+				else:
+					self.error = True
+					
+		if self.crystals is not None:
+			self.crystal_list, self.identifiers = self.retrieve_files(self.crystals)
+			#The crystal naming scheme includes an identifier at the end. This removes it.
+			self.identifiers = [id[:-13] for id in self.identifiers]
+			
+				
+		if not self.error:
+			self.prep_input_files()
+			
+	def prep_input_files(self):
+		self.in_files = []
+		if self.genome_list is not None:
+			self.status = "genome"
+			for g in self.genome_list:
+				f = input_file(g, output = self.output, do_compress = self.do_comp, make_crystal = self.crystalize)
+				f.set_genome(g)
+				self.in_files.append(f)
+			
+		if self.protein_list is not None:
+			self.status = "protein"
+			for p in self.protein_list:
+				f = input_file(p, output = self.output, do_compress = self.do_comp, make_crystal = self.crystalize)
+				f.set_protein(p)
+				self.in_files.append(f)
+			
+		if self.hmm_list is not None:
+			self.status = "protein+HMM"
+			for h, f in zip(self.hmm_list, self.in_files):
+				f.set_hmm(h)
+	
+def sql_query(genomes, proteins, hmms, db_name, output, threads, verbose, do_stdev, style, in_mem, make_db, qdb_name, do_comp):
+	
+	if not os.path.exists(db_name):
+		print("")
+		print("FastAAI can't find your database:", db_name)
+		print("Are you sure that the path you've given to the database is correct and that the database exists?")
+		print("FastAAI exiting.")
+		print("")
+		sys.exit()
+		
+	#importer opts
+	#genomes = None, proteins = None, hmms = None, crystals = None
+	imported_files = fastaai_file_importer(genomes = genomes, proteins = proteins, hmms = hmms, output = output)
+	imported_files.determine_inputs()
+
+	if imported_files.error:
+		print("Exiting FastAAI due to input file error.")
+		quit()
+
+	good_to_go = prepare_directories(output, imported_files.status, "query")
+	
+	if not good_to_go:
+		print("Exiting FastAAI")
+		sys.exit()
+	
+	print("")
+
+	'''
+	self, in_memory = False, input_file_objects = None,
+	target = None, threads = 1, do_sd = False, output_base = "FastAAI", output_style = "tsv", 
+	build_db_from_queries = True, qdb_name = "Query_FastAAI_database.db", hmm_path = "00.Libraries/01.SCG_HMMs/Complete_SCG_DB.hmm", 
+	do_comp = True, verbose = True
+	'''
+	hmm_path = find_hmm()
+	
+	mdb = file_vs_db_query(in_memory = in_mem, input_file_objects = imported_files.in_files, target=db_name, 
+	threads = threads, output_base = output, do_sd = do_stdev, output_style = style, do_comp = do_comp, 
+	build_db_from_queries = make_db, qdb_name = qdb_name, verbose = verbose, hmm_path = hmm_path)
+	
+	mdb.run()
+	
+	#Here's where the querying db comes in
+
+		
+	print("FastAAI query complete! Results at:", os.path.normpath(output + "/results"))
+	return None
+	
+#Manages the query process.
+def db_query_opts():
+	parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter,
+			description='''
+	This FastAAI module takes two FastAAI databases and searches all of the genomes in the QUERY against all of the genomes in the TARGET
+	
+	If you have many genomes (more than 1000), it will be faster to create the query database using FastAAI build_db, 
+	then search it against an existing target using this module than it is to do the same thing with an SQL query.
+	
+	If you give the same database as query and target, a special all vs. all search of the genomes in the database will be done.
+	''')
+	parser.add_argument('-q', '--query',  dest = 'query',  default = None, help =  'Path to the query database. The genomes FROM the query will be searched against the genomes in the target database')
+	parser.add_argument('-t', '--target', dest = 'target', default = None, help =  'Path to the target database.')
+				
+	parser.add_argument('-o', '--output', dest = 'output', default = "FastAAI", help = 'The directory where FastAAI will place the result of this query. By default, a directory named "FastAAI" will be created in the current working directory and results will be placed there.')
+	parser.add_argument('--output_style', dest = "style", default = 'tsv', help = "Either 'tsv' or 'matrix'. Matrix produces a simplified output of only AAI estimates.")
+	parser.add_argument('--do_stdev',   dest = "do_stdev", action='store_true',                 help = 'Off by default. Calculate std. deviations on Jaccard indicies. Increases memory usage and runtime slightly. Does NOT change estimated AAI values at all.')
+
+	parser.add_argument('--threads',      dest = 'threads', type=int, default = 1, help = 'The number of processors to use. Default 1.')
+	parser.add_argument('--verbose',      dest = 'verbose', action='store_true', help = 'Print minor updates to console. Major updates are printed regardless.')
+	parser.add_argument('--in_memory', dest = "in_mem", action = 'store_true', help = 'Load both databases into memory before querying. Consumes more RAM, but is faster and reduces file I/O substantially. Consider reducing number of threads')
+	parser.add_argument('--store_results', dest = "storage", action = 'store_true', help = 'Keep partial results in memory. Only works with --in_memory. Fewer writes, but more RAM. Default off.')
+	
+	args, unknown = parser.parse_known_args()
+	
+	return parser, args
+	
+	
+#db-db query; in-mem		
+def parse_db_init(query, target, outpath):
+	global qdb
+	qdb = query
+	global tdb
+	tdb = target
+	global output_path
+	output_path = outpath
+	
+	global query_gak
+	global target_gak
+	
+	return qdb, tdb, output_path
+	
+def parse_accession(acc):
+	tmp = sqlite3.connect(":memory:")
+	curs = tmp.cursor()
+	
+	curs.execute("attach '" + qdb + "' as queries")
+	curs.execute("attach '" + tdb + "' as targets")
+	
+	sql = '''
+	SELECT queries.{acc}.genomes, targets.{acc}.genomes 
+	FROM queries.{acc} INNER JOIN targets.{acc} 
+	ON queries.{acc}.kmer=targets.{acc}.kmer
+	'''.format(acc = acc)
+	
+	res = curs.execute(sql).fetchall()
+	
+	curs.execute("detach queries")
+	curs.execute("detach targets")
+	
+	curs.close()
+	tmp.close()
+	
+	tl = []
+	ql = {}
+	
+	acc_id = generate_accessions_index()
+	acc_id = acc_id[acc]
+	
+	indexer = 0
+	for r in res:
+		queries = np.frombuffer(r[0], dtype = np.int32)
+		tgt = np.frombuffer(r[1], dtype = np.int32)
+		tl.append(tgt)
+		
+		for q in queries:
+			if q not in ql:
+				ql[q] = {}
+			if acc_id not in ql[q]:
+				ql[q][acc_id] = []
+				
+			ql[q][acc_id].append(indexer)
+		
+		indexer += 1
+		
+	tl = np.array(tl, dtype = object)
+			
+	for q in ql:
+		if acc_id in ql[q]:
+			ql[q][acc_id] = np.array(ql[q][acc_id], dtype=np.int32)
+	
+	out_file = os.path.normpath(output_path+"/"+acc+".pickle")
+	
+	with open(out_file, "wb") as out:
+		pickle.dump([ql, tl], out)
+		
+	return([acc, out_file])
+			
+#all of this is exclusive to the in-mem approach for db db query
+def one_init(ql, tl, num_tgt, qgak_queue, tgak, tpres, sd, sty, output_dir, store_results, progress_queue, qnames, tnames, temp_dir):
+	global _ql
+	_ql = ql
+	global _tl
+	_tl = tl
+	global _nt
+	_nt = num_tgt
+	
+	qgak_data = qgak_queue.get()
+	
+	global out_base
+	out_base = output_dir
+	
+	global group_id
+	group_id = os.path.normpath(temp_dir + "/partial_results_group_" + str(qgak_data[0])+ ".txt")
+	
+	global _qgak
+	_qgak = qgak_data[1]
+	
+	global query_grouping
+	query_grouping = qgak_data[2]
+	
+	qgak_data = None
+	
+	global _tgak
+	_tgak = tgak
+	
+	global _tpres
+	_tpres = tpres
+	
+	global _tct
+	_tct = np.sum(_tpres, axis = 0)
+	
+	global do_sd
+	do_sd = sd
+	global style
+	style = sty
+	#Suppress div by zero warning - it's handled.
+	np.seterr(divide='ignore')
+	
+	global store
+	store = store_results
+	if store:
+		global holder
+		holder = []
+	else:
+		global outwriter
+		outwriter = open(group_id, "w")
+		
+	global prog_queue
+	prog_queue = progress_queue
+	
+	global _qnames
+	_qnames = qnames
+	
+	global _tnames
+	_tnames = tnames
+	
+def one_work(placeholder):	
+	for q in query_grouping:
+		results = []
+		#We also need to count the accs in the query genome, but which are not part of the inner join.
+		for acc in _qgak[q][0]:
+			if acc in _ql[q]:
+				#the bincount is intersections.
+				these_intersections = np.bincount(np.concatenate(_tl[acc][_ql[q][acc]]), minlength = _nt)
+			else:
+				#there are no intersections even though this accession is shared with at least one target
+				#number of intersects is all zeros
+				these_intersections = np.zeros(_nt, dtype = np.int32)
+			
+			#Append the counts or zeros, either way.
+			results.append(these_intersections)
+				
+		results = np.vstack(results)
+		
+		target_kmer_counts = _tgak[_qgak[q][0], :]
+		
+		#unions = size(A) + size(B) - size(intersections(A, B)) 
+		#unions = target_kmer_counts + query_kmers_by_acc - intersections
+		unions = np.subtract(np.add(target_kmer_counts, _qgak[q][1][:, None]), results)
+		
+		#These are now jaccards, not #intersections
+		results = np.divide(results, unions)
+		
+		shared_acc_counts = np.sum(_tpres[_qgak[q][0], :], axis = 0)
+		
+		no_hit = np.where(shared_acc_counts == 0)
+		
+		jaccard_averages = np.divide(np.sum(results, axis = 0), shared_acc_counts)
+
+		#Skip SD if output is matrix
+		if style == "tsv":
+			aai_ests = numpy_kaai_to_aai(jaccard_averages)
+
+			if do_sd:
+				#find diffs from means; this includes indicies corresponding to unshared SCPs that should not be included. 
+				results = results - jaccard_averages
+				
+				#fix those corresponding indicies to not contribute to the final SD.
+				results[np.logical_not(_tpres[_qgak[q][0], :])] = 0
+				#results[np.nonzero(has_accs == 0)] = 0
+				
+				#Square them; 0^2 = 0, so we don't have to think about the fixed indices any more.
+				results = np.square(results)
+				#Sum squares and divide by shared acc. count, the sqrt to get SD.
+				jaccard_SDs = np.sqrt(np.divide(np.sum(results, axis = 0), shared_acc_counts))
+				jaccard_SDs = np.round(jaccard_SDs, 4).astype(str)
+				
+			no_hit = np.where(shared_acc_counts == 0)
+			
+			#addtl.shape[0] is the query acc count
+			possible_hits = np.minimum(_qgak[q][0].shape[0], _tct).astype(str)
+			
+			jaccard_averages = np.round(jaccard_averages, 4).astype(str)
+			shared_acc_counts = shared_acc_counts.astype(str)
+			
+			jaccard_averages[no_hit] = "N/A"
+			aai_ests[no_hit] = "N/A"
+			shared_acc_counts[no_hit] = "N/A"
+			possible_hits[no_hit] = "N/A"
+				
+			qname = _qnames[q]
+				
+			output_name = os.path.normpath(out_base + "/results/"+qname+"_results.txt")
+		
+			out = open(output_name, "w")
+			out.write("query\ttarget\tavg_jacc_sim\tjacc_SD\tnum_shared_SCPs\tposs_shared_SCPs\tAAI_estimate\n")
+			if do_sd:
+				jaccard_SDs[no_hit] = "N/A"
+				for i in range(0, len(aai_ests)):
+					out.write(qname+"\t"+_tnames[i]+"\t"+jaccard_averages[i]+"\t"+jaccard_SDs[i]+"\t"+shared_acc_counts[i]+"\t"+possible_hits[i]+"\t"+aai_ests[i]+"\n")
+			else:			
+				for i in range(0, len(aai_ests)):
+					out.write(qname+"\t"+_tnames[i]+"\t"+jaccard_averages[i]+"\t"+"N/A"+"\t"+shared_acc_counts[i]+"\t"+possible_hits[i]+"\t"+aai_ests[i]+"\n")
+			out.close()
+			
+			
+		else:
+			if store:
+				aai_ests = numpy_kaai_to_aai_just_nums(jaccard_averages, as_float = False)
+				aai_ests[no_hit] = 0
+				#add zeros at misses/NAs
+				holder.append(aai_ests)
+			else:
+				aai_ests = numpy_kaai_to_aai_just_nums(jaccard_averages, as_float = True)
+				aai_ests[no_hit] = 0
+				print(*aai_ests, sep = "\t", file = outwriter)
+		
+		prog_queue.put(q)
+	
+	prog_queue.put("done")
+	
+	return None
+
+def two_work(i):
+	if store:
+		hold_together = np.vstack(holder)
+		np.savetxt(group_id, hold_together, delimiter = "\t", fmt='%4d')
+	else:
+		outwriter.close()
+		
+	return group_id
+	
+def on_disk_init(query_database_path, target_database_path, num_tgt, query_queue, target_gak, tpres, sd, sty, output_dir, progress_queue, qnames, tnames, valids, temp_dir):
+	global database
+	database = sqlite3.connect(":memory:")
+	
+	curs = database.cursor()
+	curs.execute("attach '" + query_database_path + "' as queries")
+	curs.execute("attach '" + target_database_path + "' as targets")
+	curs.close()
+	
+	global _nt
+	_nt = num_tgt
+	
+	qgak_data = query_queue.get()
+	
+	global out_base
+	out_base = output_dir
+	
+	global group_id
+	group_id = os.path.normpath(temp_dir + "/partial_results_group_" + str(qgak_data[0])+ ".txt")
+	
+	global _qgak
+	_qgak = qgak_data[1]
+	
+	global query_grouping
+	query_grouping = qgak_data[2]
+	
+	global _tgak
+	_tgak = target_gak
+	
+	global _tpres
+	_tpres = tpres
+	
+	global _tct
+	_tct = np.sum(_tpres, axis = 0)
+	
+	global do_sd
+	do_sd = sd
+	global style
+	style = sty
+	#Suppress div by zero warning - it's handled.
+	np.seterr(divide='ignore')
+	
+	if style == "matrix":
+		global outwriter
+		outwriter = open(group_id, "w")
+		
+	global prog_queue
+	prog_queue = progress_queue
+	
+	global _qnames
+	_qnames = qnames
+	
+	global _tnames
+	_tnames = tnames
+	
+	global acc_indexer
+	acc_indexer = generate_accessions_index(forward = False)
+	
+	global _valids
+	_valids = valids
+	
+def on_disk_work_one(placeholder):
+	curs = database.cursor()
+	for q in query_grouping:
+		results = []
+		qname = _qnames[q]
+		for acc in _qgak[q][0]:
+			acc_name = acc_indexer[acc]
+				
+			if acc_name in _valids:
+				
+				one = curs.execute("SELECT kmers FROM queries."+acc_name+"_genomes WHERE genome=?", (str(q),)).fetchone()[0]
+				one = np.frombuffer(one, dtype = np.int32)
+				
+				if one.shape[0] > 998:
+					#Each kmer needs to be a tuple.
+					these_kmers = [(int(kmer),) for kmer in one]
+				
+					temp_name = "_" + qname +"_" + acc_name
+					temp_name = temp_name.replace(".", "_")
+					
+					curs.execute("CREATE TEMP TABLE " + temp_name + " (kmer INTEGER)")
+					insert_table = "INSERT INTO " + temp_name + " VALUES (?)"
+					curs.executemany(insert_table, these_kmers)
+					
+					join_and_select_sql = "SELECT genomes FROM " + temp_name + " INNER JOIN targets." + acc_name + " ON "+ temp_name+".kmer = targets." + acc_name + ".kmer;"
+					
+					matches = curs.execute(join_and_select_sql).fetchall()
+				else:
+					#kmers must be a list, not a tuple.
+					these_kmers = [int(kmer) for kmer in one]
+					select = "SELECT genomes FROM targets." + acc_name + " WHERE kmer IN ({kmers})".format(kmers=','.join(['?']*len(these_kmers)))
+					matches = curs.execute(select, these_kmers).fetchall()
+				
+				set = []
+				for row in matches:
+					set.append(row[0])
+				set = b''.join(set)
+					
+				matches = None
+				these_intersections = np.bincount(np.frombuffer(set, dtype = np.int32), minlength = _nt)
+				set = None
+				results.append(these_intersections)
+				
+			else:
+				results.append(np.zeros(_nt, dtype=np.int32))
+			
+		results = np.vstack(results)
+		
+		target_kmer_counts = _tgak[_qgak[q][0], :]
+		
+		#unions = size(A) + size(B) - size(intersections(A, B)) 
+		#unions = target_kmer_counts + query_kmers_by_acc - intersections
+		unions = np.subtract(np.add(target_kmer_counts, _qgak[q][1][:, None]), results)
+		
+		#These are now jaccards, not #intersections
+		results = np.divide(results, unions)
+
+		shared_acc_counts = np.sum(_tpres[_qgak[q][0], :], axis = 0)
+		
+		no_hit = np.where(shared_acc_counts == 0)
+		
+		jaccard_averages = np.divide(np.sum(results, axis = 0), shared_acc_counts)
+		
+		#Skip SD if output is matrix
+		if style == "tsv":
+			aai_ests = numpy_kaai_to_aai(jaccard_averages)
+			
+			if do_sd:
+				#find diffs from means; this includes indicies corresponding to unshared SCPs that should not be included. 
+				results = results - jaccard_averages
+				
+				#fix those corresponding indicies to not contribute to the final SD.
+				results[np.logical_not(_tpres[_qgak[q][0], :])] = 0
+				#results[np.nonzero(has_accs == 0)] = 0
+				
+				#Square them; 0^2 = 0, so we don't have to think about the fixed indices any more.
+				results = np.square(results)
+				#Sum squares and divide by shared acc. count, the sqrt to get SD.
+				jaccard_SDs = np.sqrt(np.divide(np.sum(results, axis = 0), shared_acc_counts))
+				jaccard_SDs = np.round(jaccard_SDs, 4).astype(str)
+				
+			no_hit = np.where(shared_acc_counts == 0)
+			
+			#_qgak[q][0] is the query acc count
+			possible_hits = np.minimum(_qgak[q][0].shape[0], _tct).astype(str)
+			
+			jaccard_averages = np.round(jaccard_averages, 4).astype(str)
+			shared_acc_counts = shared_acc_counts.astype(str)
+			
+			jaccard_averages[no_hit] = "N/A"
+			aai_ests[no_hit] = "N/A"
+			shared_acc_counts[no_hit] = "N/A"
+			possible_hits[no_hit] = "N/A"
+			
+			output_name = os.path.normpath(out_base + "/results/"+qname+"_results.txt")
+		
+			out = open(output_name, "w")
+			out.write("query\ttarget\tavg_jacc_sim\tjacc_SD\tnum_shared_SCPs\tposs_shared_SCPs\tAAI_estimate\n")
+			if do_sd:
+				jaccard_SDs[no_hit] = "N/A"
+				for i in range(0, len(aai_ests)):
+					out.write(qname+"\t"+_tnames[i]+"\t"+jaccard_averages[i]+"\t"+jaccard_SDs[i]+"\t"+shared_acc_counts[i]+"\t"+possible_hits[i]+"\t"+aai_ests[i]+"\n")
+			else:			
+				for i in range(0, len(aai_ests)):
+					out.write(qname+"\t"+_tnames[i]+"\t"+jaccard_averages[i]+"\t"+"N/A"+"\t"+shared_acc_counts[i]+"\t"+possible_hits[i]+"\t"+aai_ests[i]+"\n")
+			out.close()
+			
+		else:
+			aai_ests = numpy_kaai_to_aai_just_nums(jaccard_averages, as_float = True)
+			aai_ests[no_hit] = 0
+			print(*aai_ests, sep = "\t", file = outwriter)
+		
+		prog_queue.put(q)
+		
+	curs.close()
+	prog_queue.put("done")
+	
+def on_disk_work_two(i):
+	outwriter.close()
+	return group_id
+
+def sorted_nicely(l):
+    convert = lambda text: int(text) if text.isdigit() else text 
+    alphanum_key = lambda key: [ convert(c) for c in re.split('([0-9]+)', key) ] 
+    return sorted(l, key = alphanum_key)
+	
+class db_db_remake:
+	def __init__(self, in_memory = False, store_mat_res = False, 
+		query = None, target = None, threads = 1, do_sd = False, 
+		output_base = "FastAAI", output_style = "tsv", verbose = True):
+		
+		#databases to eat
+		self.q = query
+		self.t = target
+		
+		#metadata
+		self.ok = generate_accessions_index(forward = True)
+		self.rev = generate_accessions_index(forward = False)
+		self.valids = None
+		
+		#Originally this was made to be a memory database only block of code, but just if/else one change makes it work on disk and it doesn't need a redev, then.
+		self.as_mem_db = in_memory
+		self.store_mat = store_mat_res
+		
+		#in-mem stuff
+		self.conn = None
+		self.curs = None
+		
+		self.threads = threads
+		self.do_sd = do_sd
+
+		self.output_base = output_base
+		self.output = os.path.normpath(output_base + "/results")
+		self.style = output_style
+			
+		self.query_names = None
+		self.target_names = None
+		
+		self.num_queries = None
+		self.num_targets = None
+		
+		self.query_gak = None
+		self.target_gak = None
+		self.target_presence = None
+		
+		self.query_dict = None
+		self.target_dict = None
+		
+		self.verbose = verbose
+	
+	#getting the db metadata happens the same way in every case
+	def open(self):
+		if self.verbose:
+			print("Perusing database metadata")
+			
+		self.conn = sqlite3.connect(":memory:")			
+		self.curs = self.conn.cursor()
+		
+		self.curs.execute("attach '" + self.q + "' as queries")
+		self.curs.execute("attach '" + self.t + "' as targets")
+			
+		#Find the shared accessions for these databases
+		shared_accs_sql = '''
+		SELECT queries.sqlite_master.name 
+		FROM queries.sqlite_master INNER JOIN targets.sqlite_master 
+		ON queries.sqlite_master.name = targets.sqlite_master.name
+		'''
+		self.valids = {}
+		for table in self.curs.execute(shared_accs_sql).fetchall():
+			table = table[0]
+			#Filter to 
+			if table in self.ok:
+				self.valids[table] = self.ok[table]
+		
+		self.query_names = []
+		for r in self.curs.execute("SELECT genome FROM queries.genome_index ORDER BY gen_id").fetchall():
+			self.query_names.append(r[0])
+		
+		self.target_names = []
+		for r in self.curs.execute("SELECT genome FROM targets.genome_index ORDER BY gen_id").fetchall():
+			self.target_names.append(r[0])
+			
+		self.num_queries = len(self.query_names)
+		self.num_targets = len(self.target_names)
+		
+		gak_sql = '''
+		SELECT * FROM {db}.genome_acc_kmer_counts
+		WHERE accession in ({accs})
+		ORDER BY genome
+		'''
+		
+		acc_ids = list(self.valids.values())
+		acc_ids.sort()
+		acc_ids = tuple(acc_ids)
+		
+		#query genome-acc-kmers (gak) is ordered by genome first, then accession
+		self.query_gak = {}
+		#for result in self.curs.execute(gak_sql.format(db = "queries", accs=','.join(['?']*len(self.valids))), acc_ids).fetchall():
+		for result in self.curs.execute("SELECT * FROM queries.genome_acc_kmer_counts ORDER BY genome").fetchall():
+			genome, accession, kmer_ct = result[0], result[1], result[2]
+			if genome not in self.query_gak:
+				self.query_gak[genome] = [[],[]]
+			self.query_gak[genome][0].append(accession) 
+			self.query_gak[genome][1].append(kmer_ct) 
+			
+		#refigure into numpy arrays for quicker array access later.
+		for genome in self.query_gak:
+			self.query_gak[genome] = (np.array(self.query_gak[genome][0], dtype = np.int32), np.array(self.query_gak[genome][1], dtype = np.int32))
+		
+		#Split these into ordered groups - this makes joining results at the end easier.
+		qgak_queue = multiprocessing.Queue()
+		groupings = split_seq_indices(np.arange(self.num_queries), self.threads)
+		group_id = 0
+		for group in groupings:
+			next_set = {}
+			for i in range(group[0], group[1]):
+				next_set[i] = self.query_gak[i]
+				self.query_gak[i] = None
+			#this ensures that the selection of qgak and the query index range match
+			qgak_queue.put((group_id, next_set, np.arange(group[0], group[1]),))
+			group_id += 1
+		
+		self.query_gak = qgak_queue
+		qgak_queue = None
+		
+		#tgt gak is organized by accession first, then genome
+		self.target_gak = np.zeros(shape = (122, self.num_targets), dtype = np.int32)
+		for result in self.curs.execute(gak_sql.format(db = "targets", accs=','.join(['?']*len(self.valids))), acc_ids).fetchall():
+			genome, accession, kmer_ct = result[0], result[1], result[2]
+			self.target_gak[accession, genome] += kmer_ct
+	
+		self.target_presence = self.target_gak > 0
+		self.target_presence = self.target_presence.astype(bool)
+	
+	#This needs to have a TSV write method
+	def load_in_mem(self):
+		#tempdir_path = os.path.normpath(self.output_base+"/temp")
+		tempdir_path = tempfile.mkdtemp()
+		#if not os.path.exists(tempdir_path):
+		#	os.mkdir(tempdir_path)
+			
+		ql = {}
+		tl = {}
+		for t in self.valids.values():
+			tl[t] = None
+		for i in range(0, self.num_queries):
+			ql[i] = {}
+		
+		if self.verbose:
+			tracker = progress_tracker(total = len(self.valids), message = "Loading data in memory.")
+		else:
+			print("\nLoading data in memory.")
+			
+		
+		pool = multiprocessing.Pool(self.threads, initializer = parse_db_init, 
+		initargs = (self.q, #query 
+					self.t, #target 
+					tempdir_path,)) #outpath
+		
+		for result in pool.imap_unordered(parse_accession, self.valids.keys()):
+			this_accession = result[0]
+			
+			this_acc_id = self.ok[this_accession]
+			
+			with open(result[1], "rb") as inp:
+				this_acc_data = pickle.load(inp)
+			os.remove(result[1])
+			
+			tl[this_acc_id] = this_acc_data[1]
+			
+			for q in this_acc_data[0]:
+				#We know that this acc must be in every ql for this loaded data.
+				ql[q][this_acc_id] = this_acc_data[0][q][this_acc_id]
+			if self.verbose:
+				tracker.update()
+			
+		pool.close()
+		
+		if self.verbose:
+			tracker = progress_tracker(total = self.num_queries, message = "Calculating AAI")
+		else:
+			print("\nCalculating AAI.")
+		
+		query_groups = []
+		for grouping in split_seq_indices(np.arange(self.num_queries), self.threads):
+			query_groups.append(np.arange(grouping[0], grouping[1]))
+		
+		result_queue = multiprocessing.Queue()
+		remaining_procs = self.threads
+		still_going = True
+		
+		pool = multiprocessing.Pool(self.threads, initializer = one_init, 
+		initargs = (ql, #ql 
+					tl, #tl
+					self.num_targets, #num_tgt
+					self.query_gak, #qgak_queue
+					self.target_gak, #tgak
+					self.target_presence, #tpres
+					self.do_sd, #sd
+					self.style, #sty
+					self.output_base, #output_dir
+					self.store_mat, #store_results
+					result_queue, #progress_queue
+					self.query_names, #qnames
+					self.target_names, #tnames
+					tempdir_path,)) #temp_dir
+		
+		some_results = pool.imap(one_work, query_groups)
+		
+		while still_going:
+			item = result_queue.get() 
+			if item == "done":
+				remaining_procs -= 1
+				if remaining_procs == 0:
+					still_going = False
+			else:
+				if self.verbose:
+					tracker.update()
+				else:
+					pass
+		
+		if self.style == "matrix":
+			result_files = []
+			
+			for result in pool.map(two_work, range(0, self.threads)):
+				result_files.append(result)
+			
+			pool.close()
+		
+			self.write_mat_from_files(result_files, tempdir_path)
+		else:
+			pool.close()
+
+	#This needs to be implemented from existing code.
+	def db_on_disk(self):
+		tempdir_path = tempfile.mkdtemp()
+		if self.style == "matrix":
+			self.store_mat = False
+			
+		result_queue = multiprocessing.Queue()
+		remaining_procs = self.threads
+		still_going = True
+		
+		if self.verbose:
+			tracker = progress_tracker(total = self.num_queries, message = "Calculating AAI")
+		else:
+			print("\nCalculating AAI")
+		
+		query_groups = []
+		for grouping in split_seq_indices(np.arange(self.num_queries), self.threads):
+			query_groups.append(np.arange(grouping[0], grouping[1]))
+		
+		#query_database_path, target_database_path, num_tgt, query_queue, target_gak, tpres, sd, 
+		#sty, output_dir, progress_queue, qnames, tnames, valids, temp_dir
+		pool = multiprocessing.Pool(self.threads, initializer = on_disk_init, 
+		initargs = (self.q, #query_database_path
+					self.t, #target_database_path
+					self.num_targets, #num_tgt
+					self.query_gak, #query_queue
+					self.target_gak, #target_gak
+					self.target_presence, #tpres
+					self.do_sd, #sd
+					self.style, #sty
+					self.output_base, #output_dir
+					result_queue, #progress_queue
+					self.query_names, #qnames
+					self.target_names, #tnames
+					self.valids, #valids
+					tempdir_path,)) #temp_dir
+		
+		some_results = pool.imap(on_disk_work_one, query_groups)
+		
+		while still_going:
+			item = result_queue.get() 
+			if item == "done":
+				remaining_procs -= 1
+				if remaining_procs == 0:
+					still_going = False
+			else:
+				if self.verbose:
+					tracker.update()
+				else:
+					pass
+		
+		if self.style == "matrix":
+			result_files = []
+			for result in pool.map(on_disk_work_two, range(0, self.threads)):
+				result_files.append(result)
+			
+		pool.close()
+			
+		if self.style == "matrix":
+			self.write_mat_from_files(result_files, tempdir_path)
+			
+	def write_mat_from_files(self, result_files, tempdir_path):
+		#tempdir_path = os.path.normpath(self.output_base+"/temp")
+	
+		result_files = sorted_nicely(result_files)
+		
+		#print("Combining:")
+		#for f in result_files:
+		#	print(f)
+		
+		if self.verbose:
+			tracker = progress_tracker(total = self.threads, step_size = 2, message = "Finalizing results.")
+		else:
+			print("\nFinalizing results.")
+		
+		output_file = os.path.normpath(self.output+"/FastAAI_matrix.txt")
+		final_outwriter = open(output_file, "w")
+		print("query_genome\t"+'\t'.join(self.target_names), file = final_outwriter)
+		
+		row = 0
+			
+		for f in result_files:
+			fh = open(f, "r")
+			cur = fh.readlines()
+			fh.close()
+			
+			for i in range(0, len(cur)):
+				if self.store_mat:
+					#Add the decimals - we don't need to do this is we've been writing line-wise.
+					#values will ALWAYS be 4 digits in this method, so groups of 2 dec. works.
+					cur[i] = re.sub("(\d{2})(\d{2})", "\\1.\\2", cur[i])
+				#Add in the query name to the row
+				cur[i] = self.query_names[row]+"\t"+cur[i]
+				row += 1
+			
+			final_outwriter.write(''.join(cur))
+			cur = None	
+			
+			try:
+				os.remove(f)
+			except:
+				pass
+			
+			if self.verbose:
+				tracker.update()
+		
+		final_outwriter.close()
+		
+		try:
+			if len(os.listdir(tempdir_path)) == 0:
+				shutil.rmtree(tempdir_path)
+		except:
+			pass
+		
+	def close(self):
+		self.curs.close()
+		self.curs = None
+		
+	def clean_up(self):
+		self.conn.close()
+		self.conn = None
+		
+	def run(self):
+		self.open()
+		
+		#work
+		if self.as_mem_db:
+			self.load_in_mem()
+		else:
+			self.db_on_disk()
+		
+		self.close()
+		self.clean_up()
+	
+	
+#Control the query process for any DB-first query.
+def db_query(query, target, verbose, output, threads, do_stdev, style, in_mem, store_results):
+	print("")
+	
+	#Sanity checks.
+	if target is None:
+		print("You need to supply a databasae for --target")
+		sys.exit()
+		
+	#Sanity checks.
+	if query is None:
+		print("You need to supply a databasae for --query")
+		sys.exit()
+	
+
+	
+	#Sanity checks.
+	if not os.path.exists(target):
+		print("Target database not found. Exiting FastAAI")
+		sys.exit()
+	
+	if not os.path.exists(query):
+		print("Query database not found. Exiting FastAAI")
+		sys.exit()
+		
+	#status = "exists"
+	query_ok = assess_db(query)
+	target_ok = assess_db(target)
+	
+	if query_ok != "exists":
+		print("Query database improperly formatted. Exiting FastAAI")
+		sys.exit()
+		
+	if target_ok != "exists":
+		print("Query database improperly formatted. Exiting FastAAI")
+		sys.exit()
+	
+	#Check if the database is querying against itself.
+	if target is None or query is None:
+		print("I require both a query and a target database. FastAAI exiting.")
+		sys.exit()
+	
+	if query == target:
+		print("Performing an all vs. all query on", query)
+		#all_vs_all = True
+	else:
+		print("Querying", query, "against", target)
+		#all_vs_all = False
+	
+	#Ready the output directories as needed.	
+	#The databases are already created, the only state they can be in in P+H
+	good_to_go = prepare_directories(output, "protein and HMM", "query")
+	if not good_to_go:
+		print("Exiting FastAAI")
+		sys.exit()
+	
+	#todo
+	mdb = db_db_remake(in_memory = in_mem, store_mat_res = store_results, query = query, target = target, threads = threads, do_sd = do_stdev, output_base = output, output_style = style, verbose = verbose)
+	mdb.run()
+	
+	print("")
+	
+
+#Check to see if the file exists and is a valid fastAAI db
+def assess_db(path):
+	status = None
+	if os.path.exists(path):
+		conn = sqlite3.connect(path)
+		curs = conn.cursor()
+		try:
+			sql = "SELECT name FROM sqlite_master WHERE type='table'"
+			
+			curs.row_factory = lambda cursor, row: row[0]
+			tables = curs.execute(sql).fetchall()
+			curs.row_factory = None
+			
+			curs.close()
+			conn.close()
+						
+			if len(tables) > 2 and "genome_index" in tables and "genome_acc_kmer_counts" in tables:		
+				status = "exists"
+			else:
+				status = "wrong format"
+			
+		except:
+			status = "wrong format"
+		
+	else:
+		try:
+			conn = sqlite3.connect(path)
+			conn.close()
+			status = "created"
+		except:
+			status = "unable to create"
+			
+	return status
+	
+#Add one FastAAI DB to another FastAAI DB
+def merge_db_opts():
+	parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter,
+			description='''
+	This FastAAI module allows you to add the contents of one or more FastAAI databases to another.
+	You must have at least two already-created FastAAI databases using the build_db module before this module can be used.
+	
+	Supply a comma-separated list of at least one donor database and a single recipient database. 
+	If the recipient already exists, then genomes in all the donors will be added to the recipient.
+	If the recipient does not already exist, a new database will be created, and the contents of all the donors will be added to it.
+	
+	Example:
+	FastAAI.py merge_db --donors databases/db1.db,databases/db2.db -recipient databases/db3.db --threads 3
+	This command will create a new database called "db3.db", merge the data in db1.db and db2.db, and then add the merged data into db3.db
+	
+	Only the recipient database will be modified; the donors will be left exactly as they were before running this module.
+	''')
+			
+	parser.add_argument('-d', '--donors',      dest = 'donors',    default = None,   help =  'Comma-separated string of paths to one or more donor databases. The genomes FROM the donors will be added TO the recipient and the donors will be unaltered')
+	parser.add_argument('--donor_file',  dest = 'donor_file',    default = None,   help =  'File containing paths to one or more donor databases, one per line. Use EITHER this or --donors')
+
+	parser.add_argument('-r', '--recipient',   dest = 'recipient', default = None, help =  'Path to the recipient database. Any genomes FROM the donor database not already in the recipient will be added to this database.')	
+		
+	parser.add_argument('--verbose',           dest = 'verbose',   action='store_true', help = 'Print minor updates to console. Major updates are printed regardless.')
+	
+	parser.add_argument('--threads',      dest = 'threads', type=int, default = 1, help = 'The number of processors to use. Default 1.')
+
+	args, unknown = parser.parse_known_args()
+	
+	return parser, args
+	
+def merge_db_init(indexer, table_record, donor_dbs, tempdir):
+	global mgi
+	mgi = indexer
+	global accs_per_db
+	accs_per_db = table_record
+	global tdb_list
+	tdb_list = donor_dbs
+	global work_space
+	work_space = tempdir
+	
+def acc_transformer_merge(acc_name_genomes):
+	acc_name = acc_name_genomes.split("_genomes")[0]
+	my_acc_db = os.path.normpath(work_space + "/"+acc_name+".db")
+	if os.path.exists(my_acc_db):
+		os.remove(my_acc_db)
+		
+	my_db  = sqlite3.connect(my_acc_db)
+	curs = my_db.cursor()
+	curs.execute("CREATE TABLE {acc} (kmer INTEGER PRIMARY KEY, genomes array)".format(acc=acc_name))
+	curs.execute("CREATE TABLE {acc} (genome INTEGER PRIMARY KEY, kmers array)".format(acc=acc_name_genomes))
+	my_db.commit()
+	
+	reformat = {}
+	for d in tdb_list:
+		simple_rows = []
+		#do nothing if the acc is not in the donor.
+		if acc_name_genomes in accs_per_db[d]:
+			donor_conn = sqlite3.connect(d)
+			dcurs = donor_conn.cursor()
+			data = dcurs.execute("SELECT * FROM {acc}".format(acc=acc_name_genomes)).fetchall()
+			dcurs.close()
+			donor_conn.close()
+			
+			for row in data:
+				genome, kmers = row[0], row[1]
+				new_index = mgi[d][genome]
+				#-1 is the value indicating an already-seen genome that should not be added.
+				if new_index > -1:
+					simple_rows.append((new_index, kmers,))
+					kmers = np.frombuffer(kmers, dtype=np.int32)
+					for k in kmers:
+						if k not in reformat:
+							reformat[k] = []
+						reformat[k].append(new_index)
+			
+			if len(simple_rows) > 0:
+				curs.executemany("INSERT INTO {acc} VALUES (?,?)".format(acc=acc_name_genomes), simple_rows)
+				my_db.commit()
+				
+			simple_rows = None
+			data = None
+	
+	to_add = []
+	for k in reformat:
+		as_bytes = np.array(reformat[k], dtype = np.int32)
+		as_bytes = as_bytes.tobytes()
+		reformat[k] = None
+		to_add.append((int(k), as_bytes,))
+	
+	curs.executemany("INSERT INTO {acc} VALUES (?, ?)".format(acc = acc_name), to_add)
+	
+	my_db.commit()
+	
+	to_add = None
+	
+	curs.execute("CREATE INDEX {acc}_index ON {acc} (kmer)".format(acc=acc_name))
+	my_db.commit()
+	
+	curs.close()
+	my_db.close()
+	
+	return [my_acc_db, acc_name]
+
+def merge_db(recipient, donors, donor_file, verbose, threads):
+	#Prettier on the CLI
+	if (donors is None and donor_file is None) or recipient is None:
+		print("Either donor or target not given. FastAAI is exiting.")
+		return None
+	
+	print("")
+	
+	if donors is not None:
+		donors = donors.split(",")
+
+	if donor_file is not None:
+		try:
+			donors = []
+			fh = agnostic_reader(donor_file)
+			for line in fh:
+				line = line.strip()
+				donors.append(line)
+			fh.close()
+		except:
+			sys.exit("Could not parse your donor file.")
+		
+	valid_donors = []
+	for d in donors:
+		if os.path.exists(d):
+			if d == recipient:
+				print("Donor database", d, "is the same as the recipient. This database will be skipped.")
+			else:
+				check = assess_db(d)
+				if check == "exists":
+					if d not in valid_donors:
+						valid_donors.append(d)
+					else:
+						print("It appears that database", d, "was already added to the list of donors. Did you type it twice in the list of donors? Skipping it.")
+				else:
+					if check == "created":
+						print("Donor database", d, "not found! Skipping.")
+					else:
+						print("Something was wrong with supplied database:", d+". A status check found:", check)
+		else:
+			print("Donor database", d, "not found! Are you sure the path is correct and this donor exists? This database will be skipped.")
+		
+	if len(valid_donors) == 0:
+		print("None of the supplied donor databases were able to be accessed. FastAAI cannot continue if none of these databases are valid. Exiting.")
+		sys.exit()
+		
+	recip_check = assess_db(recipient)
+		
+	if recip_check == "created" or recip_check == "exists":
+		print("Donor databases:")
+		for donor in valid_donors:
+			print("\t", donor)
+		print("Will be added to recipient database:", recipient)
+	else:
+		print("I couldn't find or create the recipient database at", recipient+".", "Does the folder you're trying to place this database in exist, and do you have permission to write files to it? FastAAI exiting.")
+		sys.exit()
+		
+	if recipient is None or len(valid_donors) == 0:
+		print("I require both a valid donor and a recipient database. FastAAI exiting.")
+		sys.exit()
+	
+	gen_counter = 0
+	multi_gen_ids = {}
+	all_gens = {}
+	
+	#Load recipient data, if any.
+	if recip_check == "exists":
+		conn = sqlite3.connect(recipient)
+		curs = conn.cursor()
+		data = curs.execute("SELECT genome, gen_id FROM genome_index").fetchall()
+		tabs = curs.execute("SELECT name FROM sqlite_master").fetchall()
+		curs.close()
+		conn.close()
+		
+		multi_gen_ids[recipient] = {}
+		for row in data:
+			genome, index = row[0], row[1]
+			all_gens[genome] = 0
+			multi_gen_ids[recipient][genome] = index
+		
+		gen_counter = max(list(multi_gen_ids[recipient].values())) + 1
+				
+	genome_index_to_add = []
+	gak_to_add = []
+	tables = {}
+	#Donors should always exist, never be created.
+	for d in valid_donors:
+		#load
+		conn = sqlite3.connect(d)
+		curs = conn.cursor()
+		data = curs.execute("SELECT * FROM genome_index").fetchall()
+		tabs = curs.execute("SELECT name FROM sqlite_master").fetchall()
+		gak = curs.execute("SELECT * FROM genome_acc_kmer_counts").fetchall()
+		curs.close()
+		conn.close()
+		multi_gen_ids[d] = {}
+		for row in data:
+			genome, index, prot_ct = row[0], row[1], row[2]
+			if genome not in all_gens:
+				all_gens[genome] = 0
+				#We need to be able to convert number to number.
+				multi_gen_ids[d][index] = gen_counter
+				genome_index_to_add.append((genome, gen_counter, prot_ct,))
+				gen_counter += 1
+			else:
+				#This is a remove condition for later.
+				multi_gen_ids[d][index] = -1
+		data = None
+		
+		for row in gak:
+			genome_id, acc_id, kmer_ct = row[0], row[1], row[2]
+			new_index = multi_gen_ids[d][genome_id]
+			if new_index > -1:
+				gak_to_add.append((new_index, acc_id, kmer_ct,))
+		
+		tables[d] = []
+		for tab in tabs:
+			tab = tab[0]
+			if tab.endswith("_genomes"):
+				tables[d].append(tab)
+		tables[d] = set(tables[d])
+	
+	all_tabs = set()
+	for t in tables:
+		all_tabs = all_tabs.union(tables[t])
+	
+	all_tabs = list(all_tabs)
+	
+	
+	temp_dir = tempfile.mkdtemp()
+	try:
+		if verbose:
+			tracker = progress_tracker(len(all_tabs), message = "Formatting data to add to database")
+		else:
+			print("Formatting data to add to database")
+		
+		conn = sqlite3.connect(recipient)
+		curs = conn.cursor()
+		
+		#indexer, table_record, donor_dbs, tempdir
+		pool = multiprocessing.Pool(threads, initializer=merge_db_init, initargs = (multi_gen_ids, tables, valid_donors, temp_dir,))
+		
+		for result in pool.imap_unordered(acc_transformer_merge, all_tabs):
+			db, accession = result[0], result[1]
+			curs.execute("CREATE TABLE IF NOT EXISTS {acc} (kmer INTEGER PRIMARY KEY, genomes array)".format(acc=accession))
+			curs.execute("CREATE TABLE IF NOT EXISTS {acc}_genomes (genome INTEGER PRIMARY KEY, kmers array)".format(acc=accession))
+			curs.execute("CREATE INDEX IF NOT EXISTS {acc}_index ON {acc}(kmer)".format(acc=accession))
+			conn.commit()
+			
+			curs.execute("attach '" + db + "' as acc")
+			conn.commit()
+			
+			#Get the genomes from worker db.
+			curs.execute("INSERT INTO {acc}_genomes SELECT * FROM acc.{acc}_genomes".format(acc=accession))
+			to_update = curs.execute("SELECT kmer, genomes, genomes FROM acc.{acc}".format(acc=accession)).fetchall()
+			update_concat_sql = "INSERT INTO {acc} VALUES (?,?) ON CONFLICT(kmer) DO UPDATE SET genomes=genomes || (?)".format(acc=accession)
+			curs.executemany(update_concat_sql, to_update)
+			conn.commit()
+			
+			curs.execute("detach acc")
+			conn.commit()
+			
+			os.remove(db)
+			
+			if verbose:
+				tracker.update()
+		
+		pool.close()
+		pool.join()
+		
+		curs.execute("CREATE TABLE IF NOT EXISTS genome_index (genome text, gen_id integer, protein_count integer)")
+		curs.execute("CREATE TABLE IF NOT EXISTS genome_acc_kmer_counts (genome integer, accession integer, count integer)")
+		
+		curs.executemany("INSERT INTO genome_index VALUES (?,?,?)", genome_index_to_add)
+		curs.executemany("INSERT INTO genome_acc_kmer_counts VALUES (?,?,?)", gak_to_add)
+		
+		curs.execute("CREATE INDEX IF NOT EXISTS kmer_acc ON genome_acc_kmer_counts (genome, accession);")
+		
+		conn.commit()
+			
+	except:
+		curs.close()
+		conn.close()
+		#Error
+		shutil.rmtree(temp_dir)
+		if recip_check == "created":
+			print("Removing created database after failure.")
+			os.remove(recipient)
+	try:
+		curs.close()
+		conn.close()
+		#Success
+		shutil.rmtree(temp_dir)
+	except:
+		pass
+		
+	print("\nDatabases merged!")
+	
+	return None
+	
+#Query 1 genome vs. 1 target using Carlos' method - just needs query, target, threads
+def single_query_opts():
+	parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter,
+			description='''
+	This FastAAI module takes a single query genome, protein, or protein and HMM pair and a single target genome, protein, or protein and HMM pair as inputs and calculates AAI between the two.
+	
+	If you supply a genome as either query or target, a protein and HMM file will be made for the genome.
+	If you supply a protein as either query or target, an HMM file will be made for it.
+	If you supply both an HMM and protein, the search will start right away. You cannot provide only an HMM.
+	
+	No database will be built, and you cannot query multiple genomes with this module.
+	
+	If you wish to query multiple genomes against themselves in all vs. all AAI search, use aai_index instead.
+	If you wish to query multiple genomes against multiple targets, use multi_query instead.
+	''')
+	parser.add_argument('-qg', '--query_genome',  dest = 'query_genome',  default = None, help =  'Query genome')
+	parser.add_argument('-tg', '--target_genome', dest = 'target_genome', default = None, help =  'Target genome')
+	
+	parser.add_argument('-qp', '--query_protein',  dest = 'query_protein',  default = None, help =  'Query protein')
+	parser.add_argument('-tp', '--target_protein', dest = 'target_protein', default = None, help =  'Target protein')
+	
+	parser.add_argument('-qh', '--query_hmm',  dest = 'query_hmm',  default = None, help =  'Query HMM')
+	parser.add_argument('-th', '--target_hmm', dest = 'target_hmm', default = None, help =  'Target HMM')
+	
+	parser.add_argument('-o', '--output', dest = 'output', default = "FastAAI", help = 'The directory where FastAAI will place the result of this query. By default, a directory named "FastAAI" will be created in the current working directory and results will be placed there.')	
+	
+	parser.add_argument('--threads',  dest = 'threads', type=int, default = 1, help = 'The number of processors to use. Default 1.')
+	parser.add_argument('--verbose',        dest = 'verbose', action='store_true', help =   'Print minor updates to console. Major updates are printed regardless.')
+	parser.add_argument('--compress', dest = "do_comp", action = 'store_true', help = 'Gzip compress generated proteins, HMMs. Off by default.')
+	
+	args, unknown = parser.parse_known_args()
+	
+	return parser, args
+
+def kaai_to_aai(kaai):
+	# Transform the kAAI into estimated AAI values
+	aai_hat = (-0.3087057 + 1.810741 * (np.exp(-(-0.2607023 * np.log(kaai))**(1/3.435))))*100
+	
+	return aai_hat
+	
+#This one's unique. It doesn't do anything with the DB, which means it doesn't access any other functionality outside of the input_file class. It just advances a pair of inputs in parallel and does intersections.
+def single_query(qf, tf, output, verbose, threads, do_compress):	
+	
+	if qf.identifiers[0] == tf.identifiers[0]:
+		print("You've selected the same query and target genome. The AAI is 100%.")
+		print("FastAAI exiting.")
+		return None
+	
+	statuses = ["genome", "protein", "protein and hmm"]
+	query_stat = statuses.index(qf.status)
+	target_stat = statuses.index(tf.status)
+	minimum_status = statuses[min(query_stat, target_stat)]
+	
+	start_printouts = ["[Genome] Protein Protein+HMM", " Genome [Protein] Protein+HMM", "Genome Protein [Protein+HMM]"]
+	
+	print("")
+	print("Query start: ", start_printouts[query_stat])
+	print("Target start:", start_printouts[target_stat])
+	print("")
+	
+	
+	qname = qf.identifiers[0]
+	tname = tf.identifiers[0]
+	
+	name = os.path.normpath(output + "/results/" + qname + "_vs_" + tname + ".aai.txt")
+	print("Output will be located at", name)
+	
+	advance_me = [qf.in_files[0], tf.in_files[0]]
+	#All we need to do this.
+	hmm_file = find_hmm()
+	pool = multiprocessing.Pool(min(threads, 2), initializer = hmm_preproc_initializer, initargs = (hmm_file, do_compress,))
+	
+	results = pool.map(run_build, advance_me)
+	
+	pool.close()
+	pool.join()
+	
+	query = results[0]
+	target = results[1]
+	
+	print(query.partial_timings())
+	print(target.partial_timings())
+	
+	#One of the printouts
+	max_poss_prots = max(len(query.best_hits_kmers), len(target.best_hits_kmers))
+	
+	accs_to_view = set(query.best_hits_kmers.keys()).intersection(set(target.best_hits_kmers.keys()))
+	
+	results = []
+	for acc in accs_to_view:
+		intersect = np.intersect1d(query.best_hits_kmers[acc], target.best_hits_kmers[acc])
+		intersect = intersect.shape[0]
+		union = query.best_hits_kmers[acc].shape[0] + target.best_hits_kmers[acc].shape[0] - intersect
+		jacc = intersect/union
+		results.append(jacc)
+		
+	results = np.array(results, dtype = np.float_)
+	
+	jacc_mean = np.mean(results)
+	jacc_std = np.std(results)
+	actual_prots = len(results)
+	poss_prots = max(len(query.best_hits_kmers), len(target.best_hits_kmers))
+	aai_est = round(kaai_to_aai(jacc_mean), 2)
+	
+	if aai_est > 90:
+		aai_est = ">90%"
+	else:
+		if aai_est < 30:
+			aai_est = "<30%"
+	
+	output = open(name, "w")
+	
+	print("query\ttarget\tavg_jacc_sim\tjacc_SD\tnum_shared_SCPs\tposs_shared_SCPs\tAAI_estimate", file = output)
+	print(qname, tname, round(jacc_mean, 4), round(jacc_std, 4), actual_prots, poss_prots, aai_est, sep = "\t", file = output)
+	
+	output.close()
+	
+	print("query\ttarget\tavg_jacc_sim\tjacc_SD\tnum_shared_SCPs\tposs_shared_SCPs\tAAI_estimate")
+	print(qname, tname, round(jacc_mean, 4), round(jacc_std, 4), actual_prots, poss_prots, aai_est, sep = "\t")
+
+	
+	print("FastAAI single query done! Estimated AAI:", aai_est)
+
+def miga_merge_opts():
+	parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter,
+			description='''
+	Hello, Miguel.
+	
+	Give one genome in nt, aa, or aa+hmm format and a database to create or add to. 
+	It'll add the genome as efficiently as possible.
+	
+	The normal merge command creates parallel processes and gathers data in 
+	one-SCP databases to add to the main DB. Great for many genomes. A lot of extra 
+	work for just one.
+	
+	This version skips the creation of subordinate DBs and just directly adds the genome. 
+	Faster, fewer writes, no parallel overhead.
+	''')
+			
+	parser.add_argument('--genome',      dest = 'gen',    default = None,   help =  'Path to one genome, FASTA format')
+	parser.add_argument('--protein',      dest = 'prot',    default = None,   help =  'Path to one protein, AA FASTA format')
+	parser.add_argument('--hmm',      dest = 'hmm',    default = None,   help =  'Path to one HMM file as predicted by FastAAI')
+
+	parser.add_argument('--output', dest = 'output', default = "FastAAI", help = 'Place the partial output files into a directory with this base. Default "FastAAI"')
+	parser.add_argument('--target',   dest = 'database', default = None, help =  'Path to the target database. The genome supplied will be added to this. The DB will be created if needed.')	
+		
+	parser.add_argument('--verbose',           dest = 'verbose',   action='store_true', help = 'Print minor updates to console. Major updates are printed regardless.')
+	parser.add_argument('--compress',           dest = 'compress',   action='store_true', help = 'Compress generated file output')
+	
+	args, unknown = parser.parse_known_args()
+	
+	return parser, args
+
+def miga_merge(infile, target_db, verbose, do_compress):
+	status = assess_db(target_db)
+	if status == "wrong format":
+		print("The database", target_db, "exists, but appears to not be a FastAAI database.")
+		print("FastAAI will not alter this file. Quitting.")
+		return None
+		
+	if status == "unable to create":
+		print("The database", target_db, "could not be created.")
+		print("Are you sure that the path you gave is valid? Quitting.")
+		return None
+
+	if verbose:
+		print("Processing genome")
+	
+	next_id = 0
+	exist_gens = {}
+	conn = sqlite3.connect(target_db)
+	curs = conn.cursor()
+	if status == 'exists':
+		for row in curs.execute("SELECT * FROM genome_index ORDER BY gen_id").fetchall():
+			genome, id, prot_ct = row[0], row[1], row[2]
+			exist_gens[genome] = id
+			next_id += 1
+			
+	if infile.basename in exist_gens:
+		print("It looks like the file you're trying to add already exists in the database.")
+		print("Adding it is too likely to corrupt the database. Quitting.")
+		return None
+
+	hmm_file = find_hmm()
+	global hmm_manager
+
+	hmm_manager = pyhmmer_manager(do_compress)
+	hmm_manager.load_hmm_from_file(hmm_file)
+	
+	infile.preprocess()
+	
+	if len(infile.best_hits_kmers) > 0:
+	
+		ok = generate_accessions_index()
+		gak_to_add = []
+		
+		gen_id = np.zeros(1, dtype = np.int32)
+		gen_id[0] = next_id
+		gen_id = gen_id.tobytes()
+		
+		for accession in infile.best_hits_kmers:
+			acc_id = ok[accession]
+			gak_to_add.append((next_id, acc_id, infile.best_hits_kmers[accession].shape[0],))
+			
+			curs.execute("CREATE TABLE IF NOT EXISTS {acc} (kmer INTEGER PRIMARY KEY, genomes array)".format(acc=accession))
+			curs.execute("CREATE TABLE IF NOT EXISTS {acc}_genomes (genome INTEGER PRIMARY KEY, kmers array)".format(acc=accession))
+			curs.execute("CREATE INDEX IF NOT EXISTS {acc}_index ON {acc}(kmer)".format(acc=accession))
+			
+			gen_first = (next_id, infile.best_hits_kmers[accession].tobytes(),)
+			curs.execute("INSERT INTO {acc}_genomes VALUES (?,?)".format(acc=accession), gen_first)
+			
+			kmers_first = []
+			for k in infile.best_hits_kmers[accession]:
+				#we know there's only one genome in these cases.
+				kmers_first.append((int(k), gen_id, gen_id, ))
+			
+			update_concat_sql = "INSERT INTO {acc} VALUES (?,?) ON CONFLICT(kmer) DO UPDATE SET genomes=genomes || (?)".format(acc=accession)
+			
+			curs.executemany(update_concat_sql, kmers_first)
+			
+		#Safety checks.
+		curs.execute("CREATE TABLE IF NOT EXISTS genome_index (genome text, gen_id integer, protein_count integer)")
+		curs.execute("CREATE TABLE IF NOT EXISTS genome_acc_kmer_counts (genome integer, accession integer, count integer)")
+		
+		gen_idx_to_add = (infile.basename, next_id, len(infile.best_hits_kmers))
+		curs.execute("INSERT INTO genome_index VALUES (?, ?, ?)", gen_idx_to_add)
+		#gak was made over the loops.
+		curs.executemany("INSERT INTO genome_acc_kmer_counts VALUES (?,?,?)", gak_to_add)
+		curs.execute("CREATE INDEX IF NOT EXISTS kmer_acc ON genome_acc_kmer_counts (genome, accession);")
+
+		conn.commit()	
+		
+	else:
+		print("No proteins to add for this genome:",infile.basename,"Database will be unaltered. Exiting.")
+	
+	curs.close()
+	conn.close()
+	
+	
+def miga_dirs(output, subdir):
+	preparation_successful = True
+	
+	if not os.path.exists(output):
+		try:
+			os.mkdir(output)
+		except:
+			print("")
+			print("FastAAI tried to make output directory: '"+ output + "' but failed.")
+			print("")
+			print("Troubleshooting:")
+			print("")
+			print("    (1) Do you have permission to create directories in the location you specified?")
+			print("    (2) Did you make sure that all directories other than", os.path.basename(output), "already exist?")
+			print("")
+			preparation_successful = False
+	
+	if preparation_successful:
+		try:
+			if not os.path.exists(os.path.normpath(output + "/" + subdir)):
+				os.mkdir(os.path.normpath(output + "/" + subdir))
+		except:
+			print("FastAAI was able to create or find", output, "but couldn't make directories there.")
+			print("")
+			print("This shouldn't happen. Do you have permission to write to that directory?")
+			
+		
+	return preparation_successful	
+	
+def miga_preproc_opts():
+	parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter,
+			description='''Build module intended for use by MiGA. 
+			
+			Performs protein prediction, HMM searching, and best hit identification, but does NOT 
+			build a database. Produces instead "crystals," which are tab-sep files containing protein, 
+			HMM accession, and original protein sequences for the best hits. These crystals can be passed
+			to "miga_db_from_crystals" action later on to rapidly create a DB from many genomes.
+	''')
+
+	parser.add_argument('-g', '--genomes',  dest = 'genomes', default = None, help =  'A directory containing genomes in FASTA format.')
+	parser.add_argument('-p', '--proteins', dest = 'proteins', default = None, help = 'A directory containing protein amino acids in FASTA format.')
+	parser.add_argument('-m', '--hmms',     dest = 'hmms', default = None, help =     'A directory containing the results of an HMM search on a set of proteins.')
+	
+	parser.add_argument('-o', '--output',   dest = 'output', default = "FastAAI", help = 'The directory to place the database and any protein or HMM files FastAAI creates. By default, a directory named "FastAAI" will be created in the current working directory and results will be placed there.')
+						
+	parser.add_argument('--threads',  dest = 'threads', type=int, default = 1, help = 'The number of processors to use. Default 1.')
+	parser.add_argument('--verbose',        dest = 'verbose', action='store_true', help = 'Print minor updates to console. Major updates are printed regardless.')
+	parser.add_argument('--compress', dest = "do_comp", action = 'store_true', help = 'Gzip compress generated proteins, HMMs. Off by default.')
+
+	args, unknown = parser.parse_known_args()
+	
+	return parser, args
+	
+def run_miga_preproc(input_file):
+	input_file.crystalize = True
+	input_file.preprocess()
+	if len(input_file.best_hits_kmers) < 1:
+		input_file.best_hits_kmers = None
+		input_file.err_log += " This file did not successfully complete. No SCPs could be found."
+		
+	return input_file 
+	
+#Produce FastAAI preprocessed files containing HMM accession and associated protein sequence
+def miga_preproc(genomes, proteins, hmms, output, threads, verbose, do_compress):
+	success = True
+	
+	imported_files = fastaai_file_importer(genomes = genomes, proteins = proteins, hmms = hmms, output = output, compress = do_compress)
+	imported_files.determine_inputs()
+	
+	if imported_files.error:
+		print("Exiting FastAAI due to input file error.")
+		quit()
+
+	#file make checks
+	p, h, c, l = True, True, True, True
+	
+	if imported_files.status == "genome":
+		p = miga_dirs(output, "predicted_proteins")
+		h = miga_dirs(output, "hmms")
+		c = miga_dirs(output, "crystals")
+		
+	if imported_files.status == "protein":
+		h = miga_dirs(output, "hmms")
+		c = miga_dirs(output, "crystals")
+	
+	if imported_files.status == "protein+HMM":	
+		c = miga_dirs(output, "crystals")
+	
+	#We always want this one.
+	l = miga_dirs(output, "logs")
+	
+	print("")
+	
+	#Check if all created directories were successful.
+	success = p and h and c and l
+	
+	if success:
+		hmm_file = find_hmm()
+	
+		if verbose:
+			tracker = progress_tracker(total = len(imported_files.in_files), message = "Processing inputs")
+		else:
+			print("Processing inputs")
+			
+		#Only build_db makes a log.
+		
+		logger = open(os.path.normpath(output+"/logs/"+"FastAAI_preprocessing_log.txt"), "a")
+		print("file", "start_date", "end_date", "starting_format", 
+		"prot_prediction_time", "trans_table", "hmm_search_time", "besthits_time", 
+		"errors", sep = "\t", file = logger)
+		
+		fail_log = open(os.path.normpath(output+"/logs/"+"FastAAI_genome_failures.txt"), "a")
+		
+		pool = multiprocessing.Pool(threads, initializer = hmm_preproc_initializer, initargs = (hmm_file, do_compress,))
+		
+		for result in pool.imap(run_miga_preproc, imported_files.in_files):
+			#log data, regardless of kind
+			print(result.basename, result.start_time, result.end_time, result.initial_state, 
+			result.prot_pred_time, result.trans_table, result.hmm_search_time, result.besthits_time,
+			result.err_log, sep = "\t", file = logger)
+			
+			if len(result.best_hits_kmers) < 1:
+				print(result.basename, file = fail_log)
+					
+			if verbose:	
+				tracker.update()
+	
+		pool.close()
+		logger.close()
+		fail_log.close()
+
+	print("FastAAI preprocessing complete!")
+		
+	return success
+
+def miga_db_from_crystals_opts():
+	parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter,
+			description='''Takes a set of crystals produced with miga_preproc and makes a database from them.
+			
+			Supply --crystals with a directory, file of paths, or list of paths just like --genomes in a build command.''')
+
+	parser.add_argument('-c', '--crystals',  dest = 'crystals', default = None, help =  'A directory containing genomes in FASTA format.')
+	parser.add_argument('-d', '--database', dest = 'db_name', default = "FastAAI_database.sqlite.db", help =  'The name of the database you wish to create or add to. The database will be created if it doesn\'t already exist and placed in the output directory. FastAAI_database.sqlite.db by default.')
+	
+	parser.add_argument('-o', '--output',   dest = 'output', default = "FastAAI", help = 'The directory to place the database and any protein or HMM files FastAAI creates. By default, a directory named "FastAAI" will be created in the current working directory and results will be placed there.')
+						
+	parser.add_argument('--threads',  dest = 'threads', type=int, default = 1, help = 'The number of processors to use. Default 1.')
+	parser.add_argument('--verbose',        dest = 'verbose', action='store_true', help = 'Print minor updates to console. Major updates are printed regardless.')
+	args, unknown = parser.parse_known_args()
+	
+	return parser, args
+	
+#This is basically a copied function, but I'm going to ignore that for now.
+def unique_kmer_miga(seq):
+	#num tetramers = len(seq) - 4 + 1, just make it -3.
+	n_kmers = len(seq) - 3
+	
+	#Converts the characters in a sequence into their ascii int value
+	as_ints = np.array([ord(i) for i in seq], dtype = np.int32)
+	
+	#create seq like 0,1,2,3; 1,2,3,4; 2,3,4,5... for each tetramer that needs a value
+	kmers = np.arange(4*n_kmers)
+	kmers = kmers % 4 + kmers // 4
+	
+	#Select the characters (as ints) corresponding to each tetramer all at once and reshape into rows of 4, 
+	#each row corresp. to a successive tetramer
+	kmers = as_ints[kmers].reshape((n_kmers, 4))
+	
+	#Given four 2-digit numbers, these multipliers work as offsets so that all digits are preserved in order when summed
+	mult = np.array([1000000, 10000, 100, 1], dtype = np.int32)
+	
+	#the fixed values effectively offset the successive chars of the tetramer by 2 positions each time; 
+	#practically, this is concatenation of numbers
+	#Matrix mult does this for all values at once.
+	return np.unique(np.dot(kmers, mult))
+
+def para_crystal_init(tdb_queue):
+	global tdb
+	global td_name
+	tdb = tdb_queue.get()
+	td_name = tdb
+	tdb = initialize_blank_db(tdb)
+	global ok
+	ok = generate_accessions_index()
+	
+def initialize_blank_db(path):
+	sqlite3.register_converter("array", convert_array)
+	worker = sqlite3.connect(path)
+	wcurs = worker.cursor()
+	wcurs.execute("CREATE TABLE genome_index (genome text, gen_id integer, protein_count integer)")
+	wcurs.execute("CREATE TABLE genome_acc_kmer_counts (genome integer, accession integer, count integer)")
+	ok = generate_accessions_index()
+	for t in ok:
+		wcurs.execute("CREATE TABLE " + t + "_genomes (genome INTEGER PRIMARY KEY, kmers array)")
+		wcurs.execute("CREATE TABLE " + t + " (kmer INTEGER PRIMARY KEY, genomes array)")
+		
+	worker.commit()
+	wcurs.close()
+	return worker
+	
+def para_crystals_to_dbs(args):
+	path, name, num = args[0], args[1], args[2]
+	my_gak = []
+	my_qgi = []
+	num_prots = 0
+	curs = tdb.cursor()
+	fh = agnostic_reader(path)
+	for line in fh:
+		segs = line.strip().split("\t")
+		#prot_name = segs[0]
+		acc_name = segs[1]
+		prot_seq = segs[2]
+		acc_id = ok[acc_name]
+		tetramers = unique_kmer_miga(prot_seq)
+		my_gak.append((num, acc_id, tetramers.shape[0]))
+		tetramers = tetramers.tobytes()
+		curs.execute("INSERT INTO " + acc_name + "_genomes VALUES (?,?)", (num, tetramers,))
+		num_prots += 1
+		
+	fh.close()
+	
+	curs.execute("INSERT INTO genome_index VALUES (?, ?, ?)", (name, num, num_prots,))
+	curs.executemany("INSERT INTO genome_acc_kmer_counts VALUES (?, ?, ?)", my_gak)
+	
+	tdb.commit()
+	curs.close()
+	
+	return None
+	
+def group_by_kmer(placeholder):
+	curs = tdb.cursor()
+	surviving_tables = []
+	for acc in ok:
+		collected_data = curs.execute("SELECT * FROM {acc}_genomes".format(acc=acc)).fetchall()
+		rearrange = {}
+		if len(collected_data) > 0:
+			surviving_tables.append(acc)
+			for row in collected_data:
+				genome, tetramers = row[0], np.frombuffer(row[1], dtype = np.int32)
+				for t in tetramers:
+					if t not in rearrange:
+						rearrange[t] = [genome]
+					else: 
+						rearrange[t].append(genome)
+			
+			to_add = []
+			for tetra in rearrange:
+				as_bytes = np.array(rearrange[tetra], dtype = np.int32).tobytes()
+				rearrange[tetra] = None
+				to_add.append((int(tetra), as_bytes,))
+				
+			curs.executemany("INSERT INTO {acc} VALUES (?, ?)".format(acc=acc), to_add)
+			to_add = None
+		else:
+			#Empty table/no genomes contained the relevant SCP
+			curs.execute("DROP TABLE {acc}".format(acc = acc))
+			curs.execute("DROP TABLE {acc}_genomes".format(acc = acc))
+			
+		tdb.commit()
+			
+	curs.close()
+	
+	tdb.close()
+	
+	return [td_name, surviving_tables]
+	
+#Merge one or many crystals into a DB.	
+def miga_db_from_crystals(crystals, output, db_name, threads, verbose):
+	success = True
+	
+	imported_files = fastaai_file_importer(genomes = None, proteins = None, 
+	hmms = None, crystals = crystals, output = output, compress = False)
+	imported_files.determine_inputs()
+	
+	if imported_files.error:
+		print("Exiting FastAAI due to input file error.")
+		quit()	
+	
+	#We'll skip trying this if the file already exists.
+	existing_genome_IDs = None
+	final_db_path = None
+	try:
+		if os.path.exists(db_name):
+			if os.path.isfile(db_name):
+				final_db_path = db_name
+			else:
+				success = miga_dirs(output, "database")
+				final_db_path = os.path.normpath(output+ "/database/" + db_name)
+				
+		else:
+			success = miga_dirs(output, "database")
+			final_db_path = os.path.normpath(output+ "/database/" + db_name)
+	except:
+		print("You specified an existing file to be a database, but it does not appear to be a FastAAI database.")
+		print("FastAAI will not be able to continue. Please give FastAAI a different database name and continue.")
+		print("Exiting.")
+		success = False
+		
+	if os.path.exists(final_db_path):
+		if os.path.isfile(final_db_path):
+			parent = sqlite3.connect(final_db_path)
+			curs = parent.cursor()
+			existing_genome_IDs = {}
+			sql_command = "SELECT genome, gen_id FROM genome_index"
+			for result in curs.execute(sql_command).fetchall():
+				genome = result[0]
+				id = int(result[1])
+				existing_genome_IDs[genome] = id
+				
+			curs.close()
+			parent.close()
+	
+	if success:
+		if existing_genome_IDs is not None:
+			genome_idx = max(list(existing_genome_IDs.values()))+1
+		else:
+			existing_genome_IDs = {}
+			genome_idx = 0
+		
+		cryst_args = []	
+		for crystal_path, crystal_name in zip(imported_files.crystal_list, imported_files.identifiers):
+			#the genome is implicitly dropped if it's already in the target
+			if crystal_name not in existing_genome_IDs:
+				existing_genome_IDs[crystal_name] = genome_idx
+				cryst_args.append((crystal_path, crystal_name, genome_idx,))
+				genome_idx += 1
+		
+		final_conn = sqlite3.connect(final_db_path)
+		final_curs = final_conn.cursor()
+		
+		final_curs.execute("CREATE TABLE IF NOT EXISTS genome_index (genome text, gen_id integer, protein_count integer)")
+		final_curs.execute("CREATE TABLE IF NOT EXISTS genome_acc_kmer_counts (genome integer, accession integer, count integer)")
+
+		final_curs.execute("CREATE INDEX IF NOT EXISTS kmer_acc ON genome_acc_kmer_counts (genome, accession);")			
+		
+		final_conn.commit()
+		
+		temp_dir = tempfile.mkdtemp()
+		
+		temp_db_queue = multiprocessing.Queue()
+		for i in range(0, threads):
+			tdb_name = os.path.normpath(temp_dir + "/temp_db_" + str(i) + ".db")
+			temp_db_queue.put(tdb_name)
+	
+		placeholder = [i for i in range(0, threads)]
+	
+		pool = multiprocessing.Pool(threads, initializer = para_crystal_init, initargs = (temp_db_queue,))
+		
+		if verbose:
+			tracker = progress_tracker(total = len(cryst_args), message = "Importing data")
+		else:
+			print("Importing data")
+		
+		for result in pool.imap_unordered(para_crystals_to_dbs, cryst_args):
+			if verbose:
+				tracker.update()
+		
+		if verbose:
+			tracker = progress_tracker(total = threads, message = "Formating data")
+		else:
+			print("Formating data")
+			
+		for result in pool.imap_unordered(group_by_kmer, placeholder):
+			dbname, surviving_tables = result[0], result[1]
+			
+			new_conn = sqlite3.connect(dbname)
+			new_curs = new_conn.cursor()
+			
+			ngak = new_curs.execute("SELECT * FROM genome_acc_kmer_counts").fetchall()
+			ngi = new_curs.execute("SELECT * FROM genome_index").fetchall()
+			
+			final_curs.executemany("INSERT INTO genome_index VALUES (?, ?, ?)", ngi)
+			final_curs.executemany("INSERT INTO genome_acc_kmer_counts VALUES (?, ?, ?)", ngak)
+			
+			final_conn.commit()
+			
+			ngak = None
+			ngi = None
+			
+			for acc in surviving_tables:
+				final_curs.execute("CREATE TABLE IF NOT EXISTS {acc}_genomes (genome INTEGER PRIMARY KEY, kmers array)".format(acc=acc))
+				final_curs.execute("CREATE TABLE IF NOT EXISTS {acc} (kmer INTEGER PRIMARY KEY, genomes array)".format(acc=acc))
+				final_curs.execute("CREATE INDEX IF NOT EXISTS {acc}_index ON {acc}(kmer)".format(acc=acc))
+
+				curag = new_curs.execute("SELECT * FROM {acc}_genomes".format(acc=acc)).fetchall()
+				final_curs.executemany("INSERT INTO {acc}_genomes VALUES (?, ?)".format(acc=acc), curag)
+				curag = None
+		
+				curaac = new_curs.execute("SELECT kmer, genomes, genomes FROM {acc}".format(acc=acc)).fetchall()
+				update_concat_sql = "INSERT INTO {acc} VALUES (?,?) ON CONFLICT(kmer) DO UPDATE SET genomes=genomes || (?)".format(acc=acc)
+				final_curs.executemany(update_concat_sql, curaac)
+				curacc = None
+					
+				final_conn.commit()
+				
+				
+				
+			new_curs.close()
+			new_conn.close()
+				
+			if verbose:
+				tracker.update()
+								
+		pool.close()
+			
+		
+		final_curs.close()
+		final_conn.close()
+		
+		shutil.rmtree(temp_dir)
+'''
+Main
+'''
+
+#Preprocess genomes, build DB, query all vs all to self.	
+def aai_index_opts():
+	parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter,
+			description='''FastAAI module for preprocessing a set of genomes, proteins, or proteins+HMMs
+			into a database, and then querying the database against itself.
+			
+			Equivalent to running build_db and db_query in sequence. Check these modules for additional
+			details on inputs.''')
+	
+	parser.add_argument('-o', '--output',   dest = 'output', default = "FastAAI", help = 'The directory to place the database and any protein or HMM files FastAAI creates. By default, a directory named "FastAAI" will be created in the current working directory and results will be placed there.')
+		
+	parser.add_argument('-g', '--genomes',  dest = 'genomes', default = None, help =  'A directory containing genomes in FASTA format.')
+	parser.add_argument('-p', '--proteins', dest = 'proteins', default = None, help = 'A directory containing protein amino acids in FASTA format.')
+	parser.add_argument('-m', '--hmms',     dest = 'hmms', default = None, help =     'A directory containing the results of an HMM search on a set of proteins.')
+							
+	parser.add_argument('-d', '--database', dest = 'db_name', default = "FastAAI_database.sqlite.db", help =  'The name of the database you wish to create or add to. The database will be created if it doesn\'t already exist and placed in the output directory. FastAAI_database.sqlite.db by default.')
+	
+	parser.add_argument('--output_style', dest = "style", default = 'tsv', help = "Either 'tsv' or 'matrix'. Matrix produces a simplified output of only AAI estimates.")
+	parser.add_argument('--do_stdev',   dest = "do_stdev", action='store_true',                 help = 'Off by default. Calculate std. deviations on Jaccard indicies. Increases memory usage and runtime slightly. Does NOT change estimated AAI values at all.')
+	parser.add_argument('--in_memory', dest = "in_mem", action = 'store_true', help = 'Load both databases into memory before querying. Consumes more RAM, but is faster and reduces file I/O substantially. Consider reducing number of threads')
+	parser.add_argument('--store_results', dest = "storage", action = 'store_true', help = 'Keep partial results in memory. Only works with --in_memory. Fewer writes, but more RAM. Default off.')
+	
+	parser.add_argument('--compress', dest = "do_comp", action = 'store_true', help = 'Gzip compress generated proteins, HMMs. Off by default.')
+	parser.add_argument('--threads',  dest = 'threads', type=int, default = 1, help = 'The number of processors to use. Default 1.')
+	parser.add_argument('--verbose',        dest = 'verbose', action='store_true', help = 'Print minor updates to console. Major updates are printed regardless.')
+					
+	args, unknown = parser.parse_known_args()
+	
+	return parser, args
+
+#Preprocess two sets of genomes A and B into two distinct databases Adb and Bdb, then query Adb against Bdb
+def multi_query_opts():
+	parser = argparse.ArgumentParser(formatter_class=argparse.RawTextHelpFormatter,
+			description='''FastAAI module for preprocessing two sets of input files into two separate DBs, 
+			then querying the DBs against eachother. Not for use with already-made FastAAI databases.
+			
+			See "build_db" action for details on file inputs.
+			See "db_query" action for details on querying options.''')
+			
+	parser.add_argument('--query_output',   dest = 'qoutput', default = "FastAAI_query", help = 'Output directory for query files. Default "FastAAI_query." FastAAI will work if this directory is the same as --target_output, but this is NOT a good idea.')
+	parser.add_argument('--target_output',   dest = 'toutput', default = "FastAAI_target", help = 'Output directory for target files. Default "FastAAI_target." AAI results will be placed in this directory')
+		
+	parser.add_argument('--query_genomes',  dest = 'qgenomes', default = None, help =  'Query genomes')
+	parser.add_argument('--target_genomes',  dest = 'tgenomes', default = None, help =  'Target genomes')
+	
+	parser.add_argument('--query_proteins', dest = 'qproteins', default = None, help = 'Query proteins')
+	parser.add_argument('--target_proteins', dest = 'tproteins', default = None, help = 'Target proteins')
+	
+	parser.add_argument('--query_hmms',     dest = 'qhmms', default = None, help =     'Query HMMs')
+	parser.add_argument('--target_hmms',     dest = 'thmms', default = None, help =     'Target HMMs')
+							
+	parser.add_argument('--query_database', dest = 'qdb_name', default = "FastAAI_query_database.sqlite.db", help =  'Query database name. Default "FastAAI_query_database.sqlite.db"')
+	parser.add_argument('--target_database', dest = 'tdb_name', default = "FastAAI_target_database.sqlite.db", help ='Target database name. Default "FastAAI_target_database.sqlite.db"')
+	
+	parser.add_argument('--output_style', dest = "style", default = 'tsv', help = "Either 'tsv' or 'matrix'. Matrix produces a simplified output of only AAI estimates.")
+	parser.add_argument('--do_stdev',   dest = "do_stdev", action='store_true',                 help = 'Off by default. Calculate std. deviations on Jaccard indicies. Increases memory usage and runtime slightly. Does NOT change estimated AAI values at all.')
+	parser.add_argument('--in_memory', dest = "in_mem", action = 'store_true', help = 'Load both databases into memory before querying. Consumes more RAM, but is faster and reduces file I/O substantially. Consider reducing number of threads')
+	parser.add_argument('--store_results', dest = "storage", action = 'store_true', help = 'Keep partial results in memory. Only works with --in_memory. Fewer writes, but more RAM. Default off.')
+	
+	parser.add_argument('--compress', dest = "do_comp", action = 'store_true', help = 'Gzip compress generated proteins, HMMs. Off by default.')
+	parser.add_argument('--threads',  dest = 'threads', type=int, default = 1, help = 'The number of processors to use. Default 1.')
+	parser.add_argument('--verbose',        dest = 'verbose', action='store_true', help = 'Print minor updates to console. Major updates are printed regardless.')
+					
+	args, unknown = parser.parse_known_args()
+	
+	return parser, args
+
+
+def main():
+	#The currently supported modules.
+	modules = ["build_db", "merge_db", "simple_query", "db_query", "single_query", "aai_index", "multi_query", "miga_merge", "miga_preproc", "miga_db_from_crystals"]
+	
+	#Print modules if someone just types FastAAI
+	if len(sys.argv) < 2:
+		print("")
+		print("    I couldn't find the module you specified. Please select one of the following modules:")
+		print("")
+		print("-------------------------------------- Database Construction Options --------------------------------------")
+		print("")
+		print("    build_db     |" + " Create or add to a FastAAI database from genomes, proteins, or proteins and HMMs")
+		print("    merge_db     |" + " Add the contents of one FastAAI DB to another")
+		print("")
+		print("---------------------------------------------- Query Options ----------------------------------------------")
+		print("")
+		print("    simple_query |" + " Query a genome or protein (one or many) against an existing FastAAI database")
+		print("    db_query     |" + " Query the genomes in one FastAAI database against the genomes in another FastAAI database")
+		print("")
+		print("------------------------------------------- Other Options -------------------------------------------")
+		print("")
+		print("    single_query |" + " Query ONE query genome against ONE target genome")
+		print("    multi_query  |" + " Create a query DB and a target DB, then calculate query vs. target AAI")
+		print("    aai_index    |" + " Create a database from multiple genomes and do an all vs. all AAI index of the genomes")
+		print("")
+		print("-----------------------------------------------------------------------------------------------------------")
+		print("    To select a module, enter 'FastAAI [module]' into the command line!")
+		print("")
+		sys.exit()
+		
+	#This is the module selection
+	selection = sys.argv[1]
+	
+	if selection == "version":
+		sys.exit("FastAAI version=0.1.17")
+	
+	if selection not in modules:
+		print("")
+		print("    I couldn't find the module you specified. Please select one of the following modules:")
+		print("")
+		print("-------------------------------------- Database Construction Options --------------------------------------")
+		print("")
+		print("    build_db     |" + " Create or add to a FastAAI database from genomes, proteins, or proteins and HMMs")
+		print("    merge_db     |" + " Add the contents of one FastAAI DB to another")
+		print("")
+		print("---------------------------------------------- Query Options ----------------------------------------------")
+		print("")
+		print("    simple_query |" + " Query a genome or protein (one or many) against an existing FastAAI database")
+		print("    db_query     |" + " Query the genomes in one FastAAI database against the genomes in another FastAAI database")
+		print("")
+		print("------------------------------------------- Other Options -------------------------------------------")
+		print("")
+		print("    single_query |" + " Query ONE query genome against ONE target genome")
+		print("    multi_query  |" + " Create a query DB and a target DB, then calculate query vs. target AAI")
+		print("    aai_index    |" + " Create a database from multiple genomes and do an all vs. all AAI index of the genomes")
+		print("")
+		print("-----------------------------------------------------------------------------------------------------------")
+		print("    To select a module, enter 'FastAAI [module]' into the command line!")
+		print("")
+		sys.exit()
+	
+#################### Database build or add ########################
+	
+	if selection == "build_db":
+		parser, opts = build_db_opts()
+		
+		#module name only
+		if len(sys.argv) < 3:
+			print(parser.print_help())
+			sys.exit()
+		
+		#Directory based
+		genomes, proteins, hmms = opts.genomes, opts.proteins, opts.hmms
+		
+		output  = os.path.normpath(opts.output)
+	
+		threads = opts.threads
+		verbose = opts.verbose
+		
+		#Database handle
+		db_name = opts.db_name
+		
+		do_comp = opts.do_comp
+		
+		build_db(genomes, proteins, hmms, db_name, output, threads, verbose, do_comp)
+	
+		
+#################### Add two DBs ########################	
+
+	if selection == "merge_db":
+		parser, opts = merge_db_opts()
+		if len(sys.argv) < 3:
+			print(parser.print_help())
+			sys.exit()
+			
+		recipient = opts.recipient
+		donors = opts.donors
+		donor_file = opts.donor_file
+		verbose = opts.verbose
+		threads = opts.threads
+		
+		if donors is not None and donor_file is not None:
+			sys.exit("You cannot specify both --donors and --donor_file.")
+		
+		merge_db(recipient, donors, donor_file, verbose, threads)
+		
+#################### Query files vs DB ########################	
+		
+	if selection == "simple_query":
+		parser, opts = sql_query_opts()
+		
+		if len(sys.argv) < 3:
+			print(parser.print_help())
+			sys.exit()
+			
+		genomes, proteins, hmms = opts.genomes, opts.proteins, opts.hmms
+		
+		db_name = opts.target
+		
+		output  = opts.output
+		threads = opts.threads
+		verbose = opts.verbose
+		
+		do_stdev = opts.do_stdev
+		
+		style, in_mem, make_db, qdb_name, do_comp = opts.style, opts.in_mem, opts.make_db, opts.qdb_name, opts.do_comp
+				
+		sql_query(genomes, proteins, hmms, db_name, output, threads, verbose, do_stdev, style, in_mem, make_db, qdb_name, do_comp)
+		
+			
+#################### Query DB vs DB ###########################		
+	if selection == "db_query":
+		parser, opts = db_query_opts()
+		#module name only
+		
+		if len(sys.argv) < 3:
+			print(parser.print_help())
+			sys.exit()
+			
+		query = opts.query
+		target = opts.target
+		verbose = opts.verbose
+		
+		do_stdev = opts.do_stdev
+		output = opts.output
+		threads = opts.threads
+		
+		style, in_mem, store = opts.style, opts.in_mem, opts.storage
+			
+		
+		db_query(query, target, verbose, output, threads, do_stdev, style, in_mem, store)
+
+#################### One-pass functions #######################
+	if selection == "single_query":
+		parser, opts = single_query_opts()
+		#module name only
+		
+		if len(sys.argv) < 3:
+			print(parser.print_help())
+			sys.exit()
+			
+		output = os.path.normpath(opts.output)
+		try:
+			threads = int(opts.threads)
+		except:
+			print("Couldn't interpret your threads. Defaulting to 1.")
+			threads = 1
+		verbose = opts.verbose
+		do_compress = opts.do_comp
+		
+		query_genome = opts.query_genome
+		query_protein = opts.query_protein
+		query_hmm = opts.query_hmm
+		
+		query_file = fastaai_file_importer(genomes = query_genome, proteins = query_protein, hmms = query_hmm, output = output, compress = do_compress)
+		query_file.determine_inputs()
+		
+		target_genome = opts.target_genome
+		target_protein = opts.target_protein
+		target_hmm = opts.target_hmm
+		
+		target_file = fastaai_file_importer(genomes = target_genome, proteins = target_protein, hmms = target_hmm, output = output, compress = do_compress)
+		target_file.determine_inputs()
+		
+		is_ok = True
+		if len(query_file.in_files) != 1:
+			print("Query genome unacceptable. Check your inputs")
+			is_ok = False
+		
+		if len(target_file.in_files) != 1:
+			print("target genome unacceptable. Check your inputs")
+			is_ok = False
+		if is_ok:
+			good_to_go = prepare_directories(output, query_file.status, "query")
+			if good_to_go:
+				good_to_go = prepare_directories(output, target_file.status, "query")
+			if good_to_go:
+				single_query(query_file, target_file, output, verbose, threads, do_compress)
+		
+		
+	if selection == "aai_index":
+		parser, opts = aai_index_opts()
+		
+		if len(sys.argv) < 3:
+			print(parser.print_help())
+			sys.exit()
+		
+		genomes, proteins, hmms = opts.genomes, opts.proteins, opts.hmms
+		
+		output  = os.path.normpath(opts.output)
+	
+		threads = opts.threads
+		verbose = opts.verbose
+		
+		#Database handle
+		db_name = opts.db_name
+		
+		do_comp = opts.do_comp
+		
+		do_stdev = opts.do_stdev
+
+		style, in_mem, store = opts.style, opts.in_mem, opts.storage
+		
+		#This is the same logic from the build_db section and it's what we need for getting the DB name.
+		#Check if the db contains path info. Incl. windows version.
+		if "/" not in db_name and "\\" not in db_name:
+			final_database = os.path.normpath(output + "/database/" + db_name)
+		else:
+			#If the person insists that the db has a path, let them.
+			final_database = db_name
+		
+		build_db(genomes, proteins, hmms, db_name, output, threads, verbose, do_comp)
+		
+		query, target = final_database, final_database
+		
+		db_query(query, target, verbose, output, threads, do_stdev, style, in_mem, store)
+		
+		
+	if selection == "multi_query":
+		parser, opts = multi_query_opts()
+		
+		if len(sys.argv) < 3:
+			print(parser.print_help())
+			sys.exit()
+		
+		#Shared options
+		threads = opts.threads
+		verbose = opts.verbose
+		
+		#query options
+		do_comp = opts.do_comp
+		do_stdev = opts.do_stdev
+		style, in_mem, store = opts.style, opts.in_mem, opts.storage
+			
+		#query inputs
+		qgenomes, qproteins, qhmms = opts.qgenomes, opts.qproteins, opts.qhmms
+		qoutput  = os.path.normpath(opts.qoutput)
+		qdb_name = opts.qdb_name
+		#This is the same logic from the build_db section and it's what we need for getting the DB name.
+		#Check if the db contains path info. Incl. windows version.
+		if "/" not in qdb_name and "\\" not in qdb_name:
+			final_qdb = os.path.normpath(qoutput + "/database/" + qdb_name)
+		else:
+			#If the person insists that the db has a path, let them.
+			final_qdb = db_name
+			
+		#target inputs
+		tgenomes, tproteins, thmms = opts.tgenomes, opts.tproteins, opts.thmms
+		toutput  = os.path.normpath(opts.toutput)
+		tdb_name = opts.tdb_name
+		#This is the same logic from the build_db section and it's what we need for getting the DB name.
+		#Check if the db contains path info. Incl. windows version.
+		if "/" not in tdb_name and "\\" not in tdb_name:
+			final_tdb = os.path.normpath(toutput + "/database/" + tdb_name)
+		else:
+			#If the person insists that the db has a path other than output/database, let them.
+			final_tdb = db_name
+		
+		#run query build
+		build_db(qgenomes, qproteins, qhmms, qdb_name, qoutput, threads, verbose, do_comp)
+		#run target build
+		build_db(tgenomes, tproteins, thmms, tdb_name, toutput, threads, verbose, do_comp)
+		#run query db against target db
+		db_query(final_qdb, final_tdb, verbose, toutput, threads, do_stdev, style, in_mem, store)
+			
+		
+############## MiGA module #################
+	if selection == "miga_merge":
+		parser, opts = miga_merge_opts()
+		
+		#module name only
+		if len(sys.argv) < 3:
+			print(parser.print_help())
+			sys.exit()
+		
+		g,p,h = opts.gen, opts.prot, opts.hmm
+		
+		target = opts.database
+		
+		verbose = opts.verbose
+		
+		output_path = opts.output
+		
+		if target == None:
+			target = os.path.normpath(output_path + "/database/FastAAI_database.sqlite.db")
+		
+		do_compress = opts.compress
+		
+		imported_files = fastaai_file_importer(genomes = g, proteins = p, hmms = h, 
+		output = output_path, compress = do_compress)
+		
+		imported_files.determine_inputs()
+		
+		if len(imported_files.in_files) == 0:
+			print("Something was wrong with your input file.")
+		else:
+			input_genome = imported_files.in_files[0]
+			
+			good_to_go = prepare_directories(output_path, imported_files.status, "build")
+			
+			miga_merge(input_genome, target, verbose, do_compress)
+		
+			#This is where a new db would normally be created, 
+			#which is not what happens when the supplied target is some other sort of path.
+			output_default = os.path.normpath(output_path + "/database")
+			if len(os.listdir(output_default)) == 0:
+				os.rmdir(output_default)
+	
+	if selection == "miga_preproc":
+		parser, opts = miga_preproc_opts()
+	
+		#module name only
+		if len(sys.argv) < 3:
+			print(parser.print_help())
+			sys.exit()
+			
+		#Directory based
+		genomes, proteins, hmms = opts.genomes, opts.proteins, opts.hmms
+		
+		output  = os.path.normpath(opts.output)
+	
+		threads = opts.threads
+		verbose = opts.verbose
+		
+		do_comp = opts.do_comp
+		
+		miga_preproc(genomes, proteins, hmms, output, threads, verbose, do_comp)
+	
+	if selection == "miga_db_from_crystals":
+		parser, opts = miga_db_from_crystals_opts()
+	
+		#module name only
+		if len(sys.argv) < 3:
+			print(parser.print_help())
+			sys.exit()
+			
+		crystals = opts.crystals
+		
+		if crystals is None:
+			print("I need to be given crystals to proceed!")
+			quit()
+		
+		db_name = opts.db_name
+		try:
+			threads = int(opts.threads)
+		except:
+			threads = 1
+			print("Can't recognize threads param:", str(opts.threads), "defaulting to 1.")
+		
+		verbose = opts.verbose
+		output_path = opts.output
+		
+		miga_db_from_crystals(crystals, output_path, db_name, threads, verbose)
+		
+	
+	return None
+	
+if __name__ == "__main__":
+	main()
+
+