bscampp 1.0.5__py3-none-any.whl → 1.0.7__py3-none-any.whl

bscampp/__init__.py

@@ -12,7 +12,7 @@ import logging, os
 # not really needed for BSCAMPP but safe to update here
 os.sys.setrecursionlimit(1000000)
 
-__version__ = "1.0.5"
+__version__ = "1.0.7"
 _INSTALL_PATH = __path__[0]
 
 # global variables to store all loggers

bscampp/configs.py

@@ -22,7 +22,6 @@ _LOG = get_logger(__name__)
 Configuration defined by users and by default values
 '''
 class Configs:
-    global _root_dir
 
     # basic input paths
     info_path = None        # info file for pplacer or EPA-ng

bscampp/functions.py

@@ -1,17 +1,84 @@
 import json, time, os, sys
 import treeswift
 from collections import defaultdict, Counter
+import subprocess
 
 from bscampp import get_logger, log_exception
 from bscampp.configs import Configs
-from bscampp.jobs import EPAngJob, TaxtasticJob, PplacerTaxtasticJob
+from bscampp.jobs import GenericJob, EPAngJob, TaxtasticJob, PplacerTaxtasticJob
 from bscampp.utils import write_fasta
 import bscampp.utils as utils
 
 import concurrent.futures
 
+# suppress userwarning when doing subtree suppress_unifurcations
+import warnings
+warnings.filterwarnings("ignore", category=UserWarning)
+
 _LOG = get_logger(__name__)
 
+############################# helper functions ################################
+'''
+Function to recompile binaries from the given directory.
+Assumption, the directory contains a CMakeLists.txt file
+'''
+def recompileBinariesFromDir(dir):
+    _LOG.warning(f"Recompiling binaries with cmake/make at {dir}")
+
+    # need to recompile the binaries
+    cmake_p = subprocess.Popen(['cmake', dir],
+            cwd=dir, stdout=subprocess.PIPE,
+            stderr=subprocess.PIPE, text=True)
+    cmake_stdout, cmake_stderr = cmake_p.communicate()
+
+    if cmake_p.returncode != 0:
+        _LOG.error("cmake failed!")
+        exit(cmake_p.returncode)
+    else:
+        _LOG.warning("cmake succeeded!")
+
+    # run make
+    make_p = subprocess.Popen(['make'],
+            cwd=dir, stdout=subprocess.PIPE,
+            stderr=subprocess.PIPE, text=True)
+    make_stdout, make_stderr = make_p.communicate()
+    
+    if make_p.returncode != 0:
+        _LOG.error(f"make failed!")
+        exit(make_p.returncode)
+    else:
+        _LOG.warning("make succeeded!")
+    _LOG.warning(f"Successfully recompiled binaries at {dir}!")
+
+'''
+Function to check hamming/fragment_hamming/homology binaries are executable, 
+since they were compiled using dynamic library
+'''
+def ensureBinaryExecutable(binpath):
+    dir = os.path.dirname(binpath)
+
+    # binpath does not exist
+    b_recompile = False
+    if not os.path.exists(binpath):
+        _LOG.warning(f"{binpath} does not exist!")
+        b_recompile = True
+    else:
+        p = subprocess.Popen([binpath], stdout=subprocess.PIPE,
+                stderr=subprocess.PIPE)
+        stdout, stderr = p.communicate()
+        # 255 or -1 indicates that the binaries work
+        if p.returncode == 255 or p.returncode == -1:
+            pass
+        else:
+            _LOG.warning(f"{binpath} return code is {p.returncode}!")
+            b_recompile = True
+
+    if b_recompile:
+        recompileBinariesFromDir(dir)
+    return
+
+########################## end of helper functions ############################
+
 '''
 Function to read in the placement tree and alignment.
 If query alignment is provided, will use the provided query instead of
@@ -47,17 +114,6 @@ def readData(workdir, dry_run=False):
         aln_dict = utils.read_data(Configs.aln_path)
         ref_dict, q_dict = utils.seperate(aln_dict, leaf_dict)
 
-    # after separating queries from the reference alignment, write
-    # them to to TEMP/
-    # Updated on 3.5.2025 by Chengze Shen
-    #   - regardless of the input choices, write a copy of both reference
-    #     and query alignment to the workdir
-    qaln_path = os.path.join(workdir, 'qaln.fa')
-    write_fasta(qaln_path, q_dict)
-    
-    aln_path = os.path.join(workdir, 'aln.fa')
-    write_fasta(aln_path, ref_dict)
-
     # Added on 3.8.2025 by Chengze Shen
     #   - to ensure that any characters from the query has correct names
     #     (e.g., having ":" can cause trouble), have a qname_map that maps
@@ -70,6 +126,22 @@ def readData(workdir, dry_run=False):
         qname_map[name] = cvt
         qname_map_rev[cvt] = name
         qidx += 1
+    # modify q_dict as well
+    for name, cvt in qname_map.items():
+        q_dict[cvt] = q_dict[name]
+        q_dict.pop(name)
+
+    # after separating queries from the reference alignment, write
+    # them to to TEMP/
+    # Updated on 3.5.2025 by Chengze Shen
+    #   - regardless of the input choices, write a copy of both reference
+    #     and query alignment to the workdir
+    qaln_path = os.path.join(workdir, 'qaln.fa')
+    write_fasta(qaln_path, q_dict)
+    
+    aln_path = os.path.join(workdir, 'aln.fa')
+    write_fasta(aln_path, ref_dict)
+
 
     t1 = time.perf_counter()
     _LOG.info('Time to read in input data: {} seconds'.format(t1 - t0))
@@ -94,18 +166,29 @@ def getClosestLeaves(aln_path, qaln_path, aln, qaln, workdir, dry_run=False):
     if Configs.subtreetype == "h":
         Configs.votes = Configs.subtreesize
 
-    cmd = []
     if Configs.similarityflag:
-        cmd.append(os.path.join(Configs.hamming_distance_dir, 'homology'))
+        job_type = 'homology'
     else:
-        if Configs.fragmentflag == False:
-            cmd.append(os.path.join(Configs.hamming_distance_dir, 'hamming'))
+        if Configs.fragmentflag:
+            job_type = 'fragment_hamming'
         else: 
-            cmd.append(os.path.join(
-                Configs.hamming_distance_dir, 'fragment_hamming'))
+            job_type = 'hamming'
+    binpath = os.path.join(Configs.hamming_distance_dir, job_type)
+    cmd = [binpath]
+
+    # Added @ 3.9.2025 by Chengze Shen
+    #   - check if binpath is executable, since the compiled files use dynamic 
+    #     libraries.
+    #     If works: should have return code 255
+    #     If not: should have return code 1,
+    #             recompile the binaries using cmake and make
+    ensureBinaryExecutable(binpath)
+
     cmd.extend([aln_path, str(len(aln)), qaln_path, str(len(qaln)),
         tmp_output, str(Configs.votes)])
-    os.system(' '.join(cmd))
+    job = GenericJob(cmd=cmd, job_type=job_type)
+    _ = job.run()
+    #os.system(' '.join(cmd))
 
     # process closest leaves 
     unusable_queries = set()
@@ -326,7 +409,8 @@ def placeQueriesToSubtrees(tree, leaf_dict, new_subtree_dict, placed_query_list,
         aln, qaln, cmdline_args, workdir, qname_map, qname_map_rev, 
         pool, lock, dry_run=False):
     t0 = time.perf_counter()
-    _LOG.info('Performing placement on each subtree...')
+    _LOG.info("Performing placement on each subtree with {}...".format(
+        Configs.placement_method))
 
     if dry_run:
         return dict()
@@ -366,10 +450,7 @@ def placeQueriesToSubtrees(tree, leaf_dict, new_subtree_dict, placed_query_list,
         if '' in tmp_leaf_dict:
             del tmp_leaf_dict['']
         tmp_ref_dict = {label : aln[label] for label in tmp_leaf_dict.keys()}
-        # Changed @ 3.8.2025 by Chengze Shen
-        #   - wrote converted name for query sequences and convert them
-        #   - back when placements are done
-        tmp_q_dict = {qname_map[name] : qaln[name] for name in query_list}
+        tmp_q_dict = {name : qaln[name] for name in query_list}
         write_fasta(tmp_aln, tmp_ref_dict)
         write_fasta(tmp_qaln, tmp_q_dict)
 
@@ -385,7 +466,7 @@ def placeQueriesToSubtrees(tree, leaf_dict, new_subtree_dict, placed_query_list,
             job = EPAngJob(path=Configs.epang_path,
                     info_path=Configs.info_path, tree_path=tmp_tree,
                     aln_path=tmp_aln, qaln_path=tmp_qaln,
-                    outdir=subtree_dir, num_cpus=Configs.num_cpus)
+                    outdir=subtree_dir, num_cpus=Configs.cpus_per_job)
             jobs.append(job)
             ## for EPA-ng, ensure that outpath name is changed to the one we want
             #_outpath = job.run(logging=f'subtree_{final_subtree_count}')
@@ -405,7 +486,7 @@ def placeQueriesToSubtrees(tree, leaf_dict, new_subtree_dict, placed_query_list,
             job = PplacerTaxtasticJob(path=Configs.pplacer_path,
                     refpkg_dir=refpkg_dir,
                     #molecule=Configs.molecule, model=Configs.model,
-                    outpath=tmp_output, num_cpus=Configs.num_cpus,
+                    outpath=tmp_output, num_cpus=Configs.cpus_per_job,
                     qaln_path=tmp_qaln)
             #tmp_output = job.run(logging=f'subtree_{final_subtree_count}')
             jobs.append(job)
@@ -445,8 +526,16 @@ def placeQueriesToSubtrees(tree, leaf_dict, new_subtree_dict, placed_query_list,
 
             for tmp_place in place_json["placements"]:
                 # convert qname back using qname_map_rev
-                qname = qname_map_rev[tmp_place[tgt][0]]
-                tmp_place[tgt][0] = qname
+                tmp_name = tmp_place[tgt][0]
+
+                # >EPA-ng: tgt=="n" --> qname is string
+                if isinstance(tmp_name, str):
+                    qname = qname_map_rev[tmp_name]
+                    tmp_place[tgt][0] = qname
+                # >pplacer: tgt=="nm" --> qname is a list of two fields
+                elif isinstance(tmp_name, list):
+                    qname = qname_map_rev[tmp_name[0]]
+                    tmp_place[tgt][0][0] = qname
                 placed_query_list.append(qname)
 
                 #placed_query_list.append(tmp_place[tgt][0])
@@ -505,6 +594,7 @@ def placeQueriesToSubtrees(tree, leaf_dict, new_subtree_dict, placed_query_list,
     _LOG.info('Time to place queries to subtrees: {} seconds'.format(t1 - t0))
     return jplace
 
+
 '''
 Function to write a given jplace object to local output
 '''

bscampp/jobs.py

@@ -112,7 +112,7 @@ class Job(object):
                 else:
                     _LOG.error(error_msg + '\nSTDOUT: ' + stdout +
                             '\nSTDERR: ' + stderr + logpath)
-                exit(1)
+                exit(self.returncode)
         except Exception:
             log_exception(_LOG)
 
@@ -123,6 +123,18 @@ class Job(object):
         raise NotImplementedError(
             'get_invocation() should be implemented by subclasses.')
 
+'''
+Generic job that runs the given command, represented as a list of strings
+'''
+class GenericJob(Job):
+    def __init__(self, cmd=[], job_type='external'):
+        Job.__init__(self)
+        self.job_type = job_type
+        self.cmd = cmd
+
+    def get_invocation(self):
+        return self.cmd, None
+
 '''
 A EPA-ng job that runs EPA-ng with given parameters
 '''

bscampp/pipeline.py

@@ -180,8 +180,6 @@ def clean_temp_files():
         _LOG.info(f'- Removed {temp}')
 
 def parseArguments(dry_run=False, method="BSCAMPP"):
-    global _root_dir, main_config_path
-
     default_outdir = f"{method.lower()}_output"
     default_outname = f"{method.lower()}_result"
 

{bscampp-1.0.5.dist-info → bscampp-1.0.7.dist-info}/METADATA

@@ -1,6 +1,6 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.4
 Name: bscampp
-Version: 1.0.5
+Version: 1.0.7
 Summary: BSCAMPP and SCAMPP - Scalable Phylogenetic Placement Tools
 Author-email: Eleanor Wedell <ewedell2@illinois.edu>, Chengze Shen <chengze5@illinois.edu>
 License: MIT License
@@ -49,12 +49,13 @@ Requires-Dist: ConfigParser>=5.0.0
 Requires-Dist: numpy>=1.21.6
 Requires-Dist: treeswift>=1.1.45
 Requires-Dist: taxtastic>=0.9.3
+Dynamic: license-file
 
 # BSCAMPP and SCAMPP - Two Scalable Phylogenetic Placement Methods and Frameworks
+[![PyPI - Version](https://img.shields.io/pypi/v/bscampp)](https://pypi.org/project/bscampp/#history)
 [![PyPI - Python Version](https://img.shields.io/pypi/pyversions/bscampp)](https://pypi.org/project/bscampp/)
-[![PyPI - Version](https://img.shields.io/pypi/v/bscampp?color=blue)](https://pypi.org/project/bscampp/#history)
 [![Build Status](https://img.shields.io/github/actions/workflow/status/ewedell/BSCAMPP/python-package.yml?branch=main&label=build)](https://github.com/ewedell/BSCAMPP/)
-[![PyPI - License](https://img.shields.io/pypi/l/bscampp?color=blue)](https://github.com/ewedell/BSCAMPP/blob/main/LICENSE)
+[![PyPI - License](https://img.shields.io/pypi/l/bscampp)](https://github.com/ewedell/BSCAMPP/blob/main/LICENSE)
 [![Changelog](https://img.shields.io/badge/CHANGELOG-grey)](https://github.com/ewedell/BSCAMPP/blob/main/CHANGELOG.md)
 
 **Table of Contents**

{bscampp-1.0.5.dist-info → bscampp-1.0.7.dist-info}/RECORD

@@ -1,10 +1,10 @@
-bscampp/__init__.py,sha256=ER9JtHb4EYnE1qyPBUmkpbsMeRC_4JDHUla46QUoInw,2289
-bscampp/configs.py,sha256=perl6u5hto6J3JV1JMbsTQ6tqr2uGOk-Z9jfzflid0s,6122
+bscampp/__init__.py,sha256=iWcT9i33I9mkOxFgfa7aRxjn_tRPYY7yW6zcuhbjTWk,2289
+bscampp/configs.py,sha256=M53nndokxi3WczhpKGWpxOcI2nXIkgpQKhj5gYpee8w,6101
 bscampp/default.config,sha256=CEfsUHBy--vwJhEcUuJ0btfuGQWb_lKMVWUIP9f5YGw,112
-bscampp/functions.py,sha256=qzlxW-bIJi0woStCzraALPb6VEPlO3CdPfCdfQqT2fQ,20119
+bscampp/functions.py,sha256=ywv3-h1l81YayioSTmNiciBZnbi56zyIGc-5Ni-jXG4,23016
 bscampp/init_configs.py,sha256=EA9sMN5jWj6zj2b-7tN19LhX2Ef61ByQLxQRLHAqLDM,3600
-bscampp/jobs.py,sha256=1FdvpSX_5VxNmJCCYAMdBKy8n68O1TjSET4XU1QULq0,7252
-bscampp/pipeline.py,sha256=IPZnXZmVxGGfbVUuGCQh5X9oBq48-6pA9QkuvMGPTag,14000
+bscampp/jobs.py,sha256=v7buZJs1AnNoXiILwu-W8fo3QjxAh3i9Mp7xfmlJvAY,7569
+bscampp/pipeline.py,sha256=J-RQH54R27m6fhzIpGX0MJuE3ZFk5rcnsROpwC_n5CE,13960
 bscampp/utils.py,sha256=-wns6FaWMKD2wVqjxdBQvjTdagTjywBIaGfqb2mupe4,30039
 bscampp/tools/epa-ng,sha256=f3EVoZAAOXLN6l521qp-TrWDl5J2nqL3tGgjPaQE9WQ,3772096
 bscampp/tools/pplacer,sha256=p0H4eo9uuiYoWS_kJbPfauOV99i7BXJdZSiwXIuLxTw,7834576
@@ -17,9 +17,9 @@ bscampp/tools/hamming_distance/src/fragment_tree_hamming.cpp,sha256=xCmyAT-OZJOD
 bscampp/tools/hamming_distance/src/fragment_tree_hamming_new.cpp,sha256=eKxgODRlpf0hU84QjNhigvRhWCT9tiJZjA5oQFQ1bUk,7404
 bscampp/tools/hamming_distance/src/homology.cpp,sha256=ZE0uXZWQ-cN4U1Wk5kUr_KKHgzsgA6Sno-IViRa4tmI,6053
 bscampp/tools/hamming_distance/src/new_hamming.cpp,sha256=fBRm99RquBZgZjaLOn9xDI3cH9NchhrxKbL-11j8fmk,5342
-bscampp-1.0.5.dist-info/LICENSE,sha256=HEa4YQdOR0e2Gz-NiOwr9X6aJcZtY0AGmlJQDmfN0Iw,1064
-bscampp-1.0.5.dist-info/METADATA,sha256=Nz0xmODp6N_e-u2VAgvegoVKW7134rM7UroUYqO7B0Q,12602
-bscampp-1.0.5.dist-info/WHEEL,sha256=5U-5D1CS1IlCq2UZGreCPlzbpvhviDLR_iCQyI6CTvY,91
-bscampp-1.0.5.dist-info/entry_points.txt,sha256=4Ft83qHc39tNNpMLgSgFXDHM-vuAB99JtmczCQj5pq8,204
-bscampp-1.0.5.dist-info/top_level.txt,sha256=1loGRUAft6Tcdq0f3lHbVwWN7W_SW1srfhAVSpg9DWE,8
-bscampp-1.0.5.dist-info/RECORD,,
+bscampp-1.0.7.dist-info/licenses/LICENSE,sha256=HEa4YQdOR0e2Gz-NiOwr9X6aJcZtY0AGmlJQDmfN0Iw,1064
+bscampp-1.0.7.dist-info/METADATA,sha256=2Tk1eF72pNPIttqSsfwiNclB8H0bShbMhca0aAr8wKo,12602
+bscampp-1.0.7.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
+bscampp-1.0.7.dist-info/entry_points.txt,sha256=4Ft83qHc39tNNpMLgSgFXDHM-vuAB99JtmczCQj5pq8,204
+bscampp-1.0.7.dist-info/top_level.txt,sha256=1loGRUAft6Tcdq0f3lHbVwWN7W_SW1srfhAVSpg9DWE,8
+bscampp-1.0.7.dist-info/RECORD,,

{bscampp-1.0.5.dist-info → bscampp-1.0.7.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (75.9.0)
+Generator: setuptools (78.1.0)
 Root-Is-Purelib: true
 Tag: py3-none-any