bacterial-annotator 0.6.6 → 0.7.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 471d9b65a3b536abb3ab252ddcf4e1b819f7dc12
-  data.tar.gz: 464f51469f99b92bc5c1717c6e0913afb8241fb0
+  metadata.gz: 0228aafd97af13b8756df42db362e4a53a30f4f0
+  data.tar.gz: 858942624597354dd0f52ad98ea1e94373289174
 SHA512:
-  metadata.gz: 7d31974ec7d7f2899366172306b33f9e456cd7827ca1f381453d3909cb16716509c66c2718af96a8e3b73f17026c6ecc0fda71b30e970e4f7aeb53fd951a7c62
-  data.tar.gz: bf7deca317fbbb5463166b9b672219875ab2fa87dcee514a5dae8bd3006a31cb563e11351c5c929bb7f10422df878788aa09e58741c11eca409af3eb0d460e1e
+  metadata.gz: 82c36c4fba00b437e721991c739517b7cfeb5edaa7e1ac49849e59d3ffac2165f1ef39f9961aa756ff8ad691fec36a8b3424cf8ce4d0e1125d486fa2e2a38593
+  data.tar.gz: e8b569f61f2dcb7309c6587ce619f7432e2588a717f3017053adcb693327ac2f21850785883eae4477226d057dde18a308a88d85d2a0558561d567865d1348cc

data/bin/ba_mash

@@ -0,0 +1,45 @@
+#!/usr/bin/env ruby
+# -*- coding: utf-8 -*-
+# author:  	maxime déraspe
+# email:	maximilien1er@gmail.com
+# review:  	
+# date:    	17-10-12
+# version: 	0.01
+# licence:  	
+
+require 'open-uri'
+
+ROOT_path = File.dirname(__FILE__)
+# url = "https://github.com/marbl/Mash/releases/download/v2.0/mash-Linux64-v2.0.tar"
+
+# Install Mash on the user system
+def installMash
+
+  begin
+    resp = open("https://github.com/marbl/Mash/releases/download/v2.0/mash-Linux64-v2.0.tar")
+    open("#{ROOT_path}/mash.tar", "wb") do |file|
+      file.write(resp.read)
+    end
+    Dir.chdir("#{ROOT_path}/")
+    `tar -xvf mash.tar > /dev/null 2>&1`
+    `cp mash-Linux64-v2.0/mash mash.linux`
+    File.chmod(0755, "#{ROOT_path}/mash.linux")
+    `rm -fr mash.tar mash-Linux64-v2.0/`
+  rescue
+    abort "Problem installing Mash, aborting"
+  end
+
+end
+
+
+# Install Mash if not already install
+if ! File.exists? "#{ROOT_path}/mash.linux"
+
+  puts "Installing Mash the MinHash sequence comparator.."
+  puts "See https://github.com/marbl/Mash/"
+  puts "The Licence is Custom"
+  installMash
+  puts "Mash successfully installed in #{ROOT_path}/mash.linux"
+  puts ""
+
+end

data/bin/bacterial-annotator

@@ -10,6 +10,7 @@
 
 require 'bacterial-annotator'
 require 'bacterial-comparator'
+require 'bacterial-identificator'
 
 VERSION = "0.6.6"
 
@@ -22,6 +23,7 @@ def print_version
   version += "  -- fasta36  v36.3.8d\n"
   version += "  -- RAxML v8.2.11\n"
   version += "  -- FastTree v2.1.10\n"
+  version += "  -- MASH v2.0\n"
   puts version
 end
 
@@ -36,6 +38,9 @@ bacterial-annotator [annotate | compare] [OPTIONS]
 
 # Choose either to annotate a genome or compare several genome annotations
 
+  identify  [OPTIONS]
+            .. see identify -h for OPTIONS
+
   annotate  [OPTIONS]
             .. see annotate -h for OPTIONS
 
@@ -64,7 +69,7 @@ annotate [OPTIONS]
     --name/-n		<name> Sample name
 
   // MERGEM-based Annotation (Recommended)
-    --db/-d          <directory> MERGEM database directory
+    --db/-d		<directory> MERGEM database directory
 
   // Reference-Based Annotation
     --refgenome/-g	<GBK_ID> 	Provide a Genbank file or a Gbk Accession ID.
@@ -122,7 +127,6 @@ def parseOptions_annotate
     when "--externaldb"
       options[:external_db] = ARGV.shift
     when "--help", "-h"
-
       usage_annotate
       abort
     when "--version", "-v"
@@ -213,6 +217,58 @@ def parseOptions_compare
 
 end
 
+
+def usage_identify
+
+  print <<OEM
+
+identify [OPTIONS] genome_1.fasta genome_2.fasta genome_x.fasta
+
+  //MERGEM Database
+    --db/-d        <database directory>
+
+  //IO
+    --proc         <nb of process> Number of process to run the comparison
+
+OEM
+
+end
+
+def parseOptions_identify
+
+  options = {}
+  options[:proc] = 2
+  options[:genomes_list] = []
+
+  while x = ARGV.shift
+
+    case x.downcase
+    when "--db", "-d"
+      options[:database] = ARGV.shift
+    when "--proc", "-p"
+      options[:proc] = ARGV.shift
+    when "--help", "-h"
+      usage_identify
+      abort
+    else
+      if File.exists? "#{x}"
+        options[:genomes_list] << x
+      else
+        puts "#{x} file doesn't exist"
+        usage_identify
+        abort
+      end
+    end
+
+  end
+
+  options
+
+end
+
+
+
+
 ########
 # MAIN #
 ########
@@ -230,9 +286,9 @@ if ARGV.size >= 1
   system("ba_fasta36")
   system("ba_cdhit")
   system("ba_fasttree")
+  system("ba_mash")
 
   options = {}
-  genomes_list = []             # TODO multiple input genomes
 
   if ARGV[0] == "annotate"
 
@@ -240,7 +296,8 @@ if ARGV.size >= 1
     options = parseOptions_annotate
 
     if ! File.exist? ("#{ROOT}/blat.linux")
-      abort "#exiting blat is missing"
+      puts "Exiting program because blat is missing"
+      abort
     end
 
     # Check Options
@@ -264,6 +321,17 @@ if ARGV.size >= 1
     bcomp.mafft_aln aln_opt
     bcomp.raxml_tree aln_opt, options[:bootstrap] if options[:phylogeny] == 1
 
+  elsif ARGV[0] == "identify"
+
+    ARGV.shift
+    options = parseOptions_identify
+    if options[:genomes_list].empty?
+      puts "You need at least 1 genome fasta to identify !!"
+      usage_identify
+      abort
+    end
+    bident = BacterialIdentificator.new(options, ROOT)
+
   elsif ARGV[0] == "--version" or ARGV[0] == "-v"
 
     print_version
@@ -273,7 +341,6 @@ if ARGV.size >= 1
     usage
   end
 
-
 else
   usage
 end

data/lib/bacterial-annotator/sequence-fasta.rb

@@ -22,8 +22,6 @@ class SequenceFasta
       abort "Aborting : The input sequence is not a fasta file !"
     end
 
-    # @contigs = extract_contigs(@fasta_flat)
-
     @meta = meta
 
     @annotation_files = nil
@@ -38,7 +36,7 @@ class SequenceFasta
 
     @annotation_files = {}
     Dir.mkdir "#{outdir}" if ! Dir.exists? "#{outdir}"
-    if @meta
+    if @meta==1
       system("#{root}/prodigal.linux -p meta -i #{@fasta_file} -a #{outdir}/Proteins.fa -d #{outdir}/Genes.fa -o #{outdir}/Genbanks.gbk -q")
     else
       system("#{root}/prodigal.linux -i #{@fasta_file} -a #{outdir}/Proteins.fa -d #{outdir}/Genes.fa -o #{outdir}/Genbanks.gbk -q")

data/lib/bacterial-annotator/sequence-synteny.rb

@@ -100,7 +100,7 @@ class SequenceSynteny
         # first hit for query
         if ! @query_sequences[key].has_key? :homology
           @query_sequences[key][:conserved] = true
-          @subject_sequences[key][:conserved] = true
+          # @subject_sequences[key][:conserved] = true
           @query_sequences[key][:homology] = {
             pId: lA[2].to_f.round(2),
             cov_query: cov_query,
@@ -119,7 +119,7 @@ class SequenceSynteny
         # query already got at least 1 hit and new_score > last_score
         elsif lA[11].to_f > @query_sequences[key][:homology][:score]
           @query_sequences[key][:conserved] = true
-          @subject_sequences[key][:conserved] = true
+          # @subject_sequences[key][:conserved] = true
           @query_sequences[key][:homology] = {
             pId: lA[2].to_f.round(2),
             cov_query: cov_query,

data/lib/bacterial-annotator.rb

@@ -26,7 +26,7 @@ class BacterialAnnotator
     @minlength = @options[:minlength].to_i
     @options[:minlength] = @options[:minlength].to_i
     @options[:pidentity] = @options[:pidentity].to_f
-    @options[:pidentispacemacs-lightty] = @options[:pidentity] * 100 if @options[:pidentity] <= 1.00
+    @options[:pidentity] = @options[:pidentity] * 100 if @options[:pidentity] <= 1.00
     @options[:pcoverage] = @options[:pcoverage].to_f
     @options[:pcoverage] = @options[:pcoverage] / 100 if @options[:pcoverage] > 1.00
 
@@ -200,8 +200,12 @@ class BacterialAnnotator
                                                 "Prot-ExternalDB", @options[:pidentity],
                                                 @options[:pcoverage], "prot")
 
-      puts "# Running BLAT alignment with External Database.."
+      print "# Running BLAT alignment with External Database.."
+      start_time = Time.now
       @externaldb_synteny.run_blat @root, @options[:outdir]
+      end_time = Time.now
+      c_time = Helper.sec2str(end_time-start_time)
+      print "done (#{c_time})\n"
       @externaldb_synteny.extract_hits :externaldb
 
       @externaldb_synteny.query_sequences.each do |k, v|
@@ -214,6 +218,9 @@ class BacterialAnnotator
 
         next if ! v.has_key? :homology
 
+        if ! @contig_annotations_cds.has_key? contig_of_protein
+          @contig_annotations_cds[contig_of_protein] = []
+        end
         @contig_annotations_cds[contig_of_protein] << k
 
         hit_gi = v[:homology][:hits][0]
@@ -238,7 +245,6 @@ class BacterialAnnotator
           inference: inference
         }
 
-
       end
 
     end
@@ -256,12 +262,17 @@ class BacterialAnnotator
       gbk_path = @query_fasta.annotation_files[:gbk_path]
       gbk_to_annotate = SequenceAnnotation.new("#{gbk_path}/#{contig}.gbk", "#{gbk_path}")
 
-      if @with_external_db
+      if @with_external_db and @with_refence_genome
         gbk_to_annotate.add_annotation_ref_synteny_prot(
           (@prot_synteny_refgenome.query_sequences.merge(@externaldb_synteny.query_sequences)),
           @contig_annotations_externaldb[contig].merge(@ref_genome.coding_seq),
           (File.basename @options[:refgenome]).gsub(/.gb.*/,"")
         )
+      elsif @with_external_db
+        gbk_to_annotate.add_annotation_ref_synteny_prot(
+          @externaldb_synteny.query_sequences,
+          @contig_annotations_externaldb[contig]
+        )
       else
         gbk_to_annotate.add_annotation_ref_synteny_prot(
           @prot_synteny_refgenome.query_sequences,
@@ -270,7 +281,7 @@ class BacterialAnnotator
         )
       end
 
-      if @contig_annotations_rna.has_key? contig
+      if @contig_annotations_rna and @contig_annotations_rna.has_key? contig
         # puts "RNA annotation"
         gbk_to_annotate.add_annotations @contig_annotations_rna[contig], "new"
       end
@@ -419,6 +430,8 @@ class BacterialAnnotator
     # >sp|C7C422|BLAN1_KLEPN Beta-lactamase NDM-1 OS=Klebsiella pneumoniae GN=blaNDM-1 PE=1 SV=1
     # TrEMBL
     # >tr|E5KIY2|E5KIY2_ECOLX Beta-lactamase NDM-1 OS=Escherichia coli GN=blaNDM-1 PE=1 SV=1
+    # MERGEM
+    # >Genome_ID|location|Protein_ID|LocusTag|Gene|Protein_Product
 
     ref_cds = {}
 
@@ -436,7 +449,10 @@ class BacterialAnnotator
           product = ""
           db_source = "[DBSource]"
 
-          if product_long.include? " [" and product_long.include? "]" # NCBI
+          if product_long.scan(/|/).count >= 5 # MERGEM
+            product = product_long
+            db_source = "RefSeq"
+          elsif product_long.include? " [" and product_long.include? "]" # NCBI
             organism = product_long[/\[.*?\]/]
             product = product_long.split(" [")[0].strip
           elsif product_long.include? "OS=" # Swissprot / TrEMBL
@@ -465,6 +481,9 @@ class BacterialAnnotator
               db_source = "UniProtKB"
             end
             prot_id = lA[1]
+          elsif key_gi.count("|") == 5
+            db_source = "RefSeq"
+            prot_id = lA[2]
           end
 
           ref_cds[key_gi] = {product: product, org: org, prot_id: prot_id, db_source: db_source}

data/lib/bacterial-comparator.rb

@@ -153,7 +153,11 @@ class BacterialComparator
     stats = {}
     stats[:syntenic] = []
     fout = File.open("#{@outdir}/cds-synteny.tsv", "w")
-    fout.write("Gene\t"+@genomes_list.join("\t")+"\n")
+    genomes_name = []
+    @genomes_list.each do |g|
+      genomes_name.push(File.basename(g))
+    end
+    fout.write("Gene\t"+genomes_name.join("\t")+"\n")
 
     to_build_multifasta = []
 
@@ -386,7 +390,6 @@ class BacterialComparator
 
   end
 
-
   def raxml_tree_dna bt
     print "# Genes DNA tree creation (RAXML).."
     start_time = Time.now
@@ -423,7 +426,6 @@ class BacterialComparator
     print "done (#{c_time})\n"
   end
 
-
   def raxml_tree aln_opt, bt
 
     if aln_opt == "both"

data/lib/bacterial-identificator.rb

@@ -0,0 +1,64 @@
+# -*- coding: utf-8 -*-
+# author:  	maxime déraspe
+# email:	maximilien1er@gmail.com
+# review:  	
+# date:    	17-10-12
+# version: 	0.0.1
+# licence:  	
+
+require 'bio'
+require 'fileutils'
+require 'parallel'
+require 'helper'
+
+class BacterialIdentificator
+
+  attr_reader :genomes_list, :stats
+
+  # Initialize BacterialIdentificator
+  # options[:input], options[:refgenome], ROOT, options[:outdir], options)
+  def initialize options, root
+
+    @root = root
+    @db_path = options[:database]
+    @genomes_list = options[:genomes_list]
+    @proc = options[:proc].to_i
+    p @genomes_list
+
+    @genomes_list.each do |g|
+      mash_genome g
+    end
+
+  end
+
+
+  def mash_genome genome
+
+    # Reference-ID, Query-ID, Mash-distance, P-value, and Matching-hashes
+    # fields = ["hit","query","distance","pvalue","match"]
+
+    results_raw = `#{@root}/mash.linux dist #{@db_path}/species-sequences.msh #{genome}`
+    results = []
+
+    results_raw.split("\n").each do |l|
+      lA = l.chomp.split("\t")
+      next if lA[-1].split("/")[0] == '0' # no match
+      results << lA
+    end
+
+    results_sorted = results.sort {|a,b| a[2] <=> b[2]}
+
+    File.open("#{genome}.msh_dist", "w") do |fout|
+      results_sorted.each do |f|
+        fout.write(f.join("\t"))
+        fout.write("\n")
+      end
+    end
+
+  end
+
+
+
+
+end
+

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: bacterial-annotator
 version: !ruby/object:Gem::Version
-  version: 0.6.6
+  version: 0.7.0
 platform: ruby
 authors:
 - Maxime Deraspe
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2017-10-10 00:00:00.000000000 Z
+date: 2017-11-06 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bio
@@ -83,6 +83,7 @@ executables:
 - ba_fasta36
 - ba_cdhit
 - ba_fasttree
+- ba_mash
 extensions: []
 extra_rdoc_files: []
 files:
@@ -92,6 +93,7 @@ files:
 - bin/ba_fasta36
 - bin/ba_fasttree
 - bin/ba_mafft
+- bin/ba_mash
 - bin/ba_prodigal
 - bin/ba_raxml
 - bin/bacterial-annotator
@@ -100,6 +102,7 @@ files:
 - lib/bacterial-annotator/sequence-fasta.rb
 - lib/bacterial-annotator/sequence-synteny.rb
 - lib/bacterial-comparator.rb
+- lib/bacterial-identificator.rb
 - lib/helper.rb
 homepage: http://rubygems.org/gems/bacterial-annotator
 licenses: