bacterial-annotator 0.7.1 → 0.8.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 10f3d2469fb3aaf64e6b84076e05ab9e1ae41cd6
-  data.tar.gz: f08a5465ce584dd888074c7d0146c1450386598e
+  metadata.gz: 1c90c64fd59297a06dd9b50def86f8ba9305b23e
+  data.tar.gz: '008eec9088d41505c8c9ffbbc097ce4cee5b5c3b'
 SHA512:
-  metadata.gz: bd006cf021f0a74f1e98fa6367ca4aca0abb36004f375654ec552b68e1ac8ebc5c1f65e38a480473551848d25ac6be904c5d1841cc60657a47384169d368a18c
-  data.tar.gz: b5a8cb5c74c028e813bbc585e70b6dcb420b8c8f4ad659e8e4c985bce868009a7f5d6015c4e396768a6146c918943e4586a638c084d844cc91be6ac927c993b6
+  metadata.gz: 7904dc06b85e1dca28bab9fbb5a84d2919da714990775df637ed27f504f25218eeffd264146fd1082e2a44e1188b2faf6e2a0ce923183eb88caf15595cf26500
+  data.tar.gz: 1c3a09f9952e5750faad5fa5c5c27492967dcebcb6f2fa5a8fd89ef79f5a074d2e6e4f4eeb299729304328777af8e752a1747409ee24150ce1f96354243a0f38

data/bin/bacterial-annotator

@@ -12,7 +12,7 @@ require 'bacterial-annotator'
 require 'bacterial-comparator'
 require 'bacterial-identificator'
 
-VERSION = "0.6.6"
+VERSION = "0.8.0"
 
 def print_version
   version = "Bacterial Annotator - Version #{VERSION}\n\n"
@@ -189,6 +189,7 @@ def parseOptions_compare
   options[:genomes_list] = []
   options[:concat] = 0
   options[:phylogeny] = 0
+  options[:software] = "fasttree"
   options[:bootstrap] = 100
 
   while x = ARGV.shift
@@ -208,6 +209,8 @@ def parseOptions_compare
       options[:concat] = ARGV.shift
     when "--phylogeny"
       options[:phylogeny] = 1
+    when "--software"
+      options[:software] = ARGV.shift
     when "--bootstrap"
       options[:bootstrap] = (ARGV.shift).to_i
     when "--help", "-h"

data/lib/bacterial-annotator.rb

@@ -47,7 +47,7 @@ class BacterialAnnotator
     Dir.mkdir(@options[:outdir])
 
     @query_fasta = SequenceFasta.new(@root,
-                                     options[:outdir],
+                                     @options[:outdir],
                                      @options[:input],
                                      @options[:meta])
 

data/lib/bacterial-comparator.rb

@@ -20,11 +20,17 @@ class BacterialComparator
   def initialize options, root
 
     @root = root
-    @outdir = options[:outdir]
+    @outdir = File.expand_path(File.dirname(options[:outdir]))+"/#{options[:outdir]}"
     Dir.mkdir(@outdir) if ! Dir.exists? @outdir
     @genomes_list = options[:genomes_list]
     @proc = options[:proc].to_i
     @phylo_nb_genes = options[:phylo_nb_genes]
+    if ["fasttree","raxml"].include? options[:software]
+      @software = options[:software]
+    else
+      @software = "fasttree"
+    end
+
     min_cov = options[:min_cov].to_f
     min_pid = options[:pidentity].to_f
     if min_cov > 1
@@ -51,7 +57,8 @@ class BacterialComparator
   def run_comparison
 
     run_mafft_aln
-    run_raxml_phylo if @run_phylo != 0
+
+    run_phylo if @run_phylo != 0
 
   end
 
@@ -119,7 +126,7 @@ class BacterialComparator
 
   def build_multifasta synteny_list
 
-    pep_out_dir = "./#{@outdir}/align-genes-pep"
+    pep_out_dir = "#{@outdir}/align-genes-pep"
 
     ref_proteins = load_genome_cds(Dir["#{@genomes_list[0]}/*.pep"][0])
     synteny_list.each do |k,v|
@@ -139,7 +146,7 @@ class BacterialComparator
 
     end
 
-    dna_out_dir = "./#{@outdir}/align-genes-dna"
+    dna_out_dir = "#{@outdir}/align-genes-dna"
     ref_genes = load_genome_cds(Dir["#{@genomes_list[0]}/*.dna"][0])
     synteny_list.each do |k,v|
       dna_out = File.open(dna_out_dir+"/#{k}.dna", "w")
@@ -220,8 +227,8 @@ class BacterialComparator
 
     fout.close
 
-    pep_out_dir = "./#{@outdir}/align-genes-pep"
-    dna_out_dir = "./#{@outdir}/align-genes-dna"
+    pep_out_dir = "#{@outdir}/align-genes-pep"
+    dna_out_dir = "#{@outdir}/align-genes-dna"
     Dir.mkdir(pep_out_dir) if ! Dir.exists? pep_out_dir
     Dir.mkdir(dna_out_dir) if ! Dir.exists? dna_out_dir
 
@@ -296,9 +303,11 @@ class BacterialComparator
     puts "# Sequence alignments - individual proteins a.a. (MAFFT)  [DONE] (in #{c_time})"
 
     # FIXME ugly hack to find out the reference genome
-    ref_id = Dir["#{ori_dir}/#{@genomes_list[0]}/*.pep"][0].split('/')[-1].gsub(".pep","")
+    Dir.chdir("#{@outdir}")
+    Dir.chdir("../")
+    ref_id = Dir["#{@genomes_list[0]}/*.pep"][0].split('/')[-1].gsub(".pep","")
 
-    concat_alignments "align-genes-pep.all.fasta", ref_id
+    concat_alignments "#{@outdir}/align-genes-pep.all.fasta", ref_id
 
     Dir.chdir(ori_dir)
 
@@ -330,13 +339,17 @@ class BacterialComparator
     end
 
     # ugly hack to find out the reference genome FIXME
-    ref_id = Dir["#{ori_dir}/#{@genomes_list[0]}/*.pep"][0].split('/')[-1].gsub(".pep","")
+    Dir.chdir("#{@outdir}")
+    Dir.chdir("../")
+
+    ref_id = Dir["#{@genomes_list[0]}/*.pep"][0].split('/')[-1].gsub(".pep","")
+    # ref_id = Dir["#{ori_dir}/#{@genomes_list[0]}/*.pep"][0].split('/')[-1].gsub(".pep","")
 
     end_time = Time.now
     c_time = Helper.sec2str(end_time-start_time)
     puts "# Sequence alignments - individual genes dna (MAFFT)  [DONE] (in #{c_time})"
 
-    concat_alignments "align-genes-dna.all.fasta", ref_id
+    concat_alignments "#{@outdir}/align-genes-dna.all.fasta", ref_id
 
     Dir.chdir(ori_dir)
 
@@ -345,15 +358,17 @@ class BacterialComparator
 
   def concat_alignments outfile, ref_id
 
-    if File.exists?("../#{outfile}") and File.size("../#{outfile}") > 0
+    if File.exists?("#{outfile}") and File.size("#{outfile}") > 0
       puts "..Alignment concatenated file already exists, skipping."
       return
     end
 
-    fout = File.open("../#{outfile}", "w")
+    fout = File.open("#{outfile}", "w")
 
     seq = ""
-    Dir["*.aln"].each do |f|
+    aln_dir = outfile.split(".")[0..-3].join(".")
+
+    Dir["#{aln_dir}/*.aln"].each do |f|
       flat = Bio::FlatFile.auto(f)
       ref_seq = flat.entries[0]
       flat.close
@@ -366,7 +381,7 @@ class BacterialComparator
 
     for i in 1..@genomes_list.length
       seq = ""
-      Dir["*.aln"].each do |f|
+      Dir["#{aln_dir}/*.aln"].each do |f|
         flat = Bio::FlatFile.auto(f)
         j=0
         flat.each_entry do |entry|
@@ -442,15 +457,57 @@ class BacterialComparator
     puts "# Proteins AA tree creation (RAXML)  [DONE] (in #{c_time})"
   end
 
-  def run_raxml_phylo
 
-    if @aln_opt == "both"
-      raxml_tree_dna @bootstrap
-      raxml_tree_pep @bootstrap
-    elsif @aln_opt == "prot"
-      raxml_tree_pep @bootstrap
-    elsif @aln_opt == "dna"
-      raxml_tree_dna @bootstrap
+  def fasttree_tree_dna bt
+    puts "# Genes DNA tree creation (FastTree)  START.."
+    start_time = Time.now
+    ori_dir = Dir.pwd
+    Dir.chdir(@outdir)
+    Dir.mkdir("tree-genes-dna") if ! Dir.exists?("tree-genes-dna")
+    current_dir = Dir.pwd
+    cmd = system("export OMP_NUM_THREADS=#{@proc} && #{@root}/fasttree.linux -nosupport -fastest -nt -gtr align-genes-dna.all.fasta > tree-genes-dna.nwk")
+    Dir.chdir(ori_dir)
+    end_time = Time.now
+    c_time = Helper.sec2str(end_time-start_time)
+    puts "# Genes DNA tree creation (FastTree)  [DONE] (in #{c_time})"
+  end
+
+
+  def fasttree_tree_pep bt
+    puts "# Proteins AA tree creation (FastTree)  START.."
+    start_time = Time.now
+    ori_dir = Dir.pwd
+    Dir.chdir(@outdir)
+    Dir.mkdir("tree-genes-pep") if ! Dir.exists?("tree-genes-pep")
+    current_dir = Dir.pwd
+    cmd = system("export OMP_NUM_THREADS=#{@proc} && #{@root}/fasttree.linux -nosupport -fastest align-genes-pep.all.fasta > tree-proteins-aa.nwk")
+    Dir.chdir(ori_dir)
+    end_time = Time.now
+    c_time = Helper.sec2str(end_time-start_time)
+    puts "# Proteins AA tree creation (FastTree)  [DONE] (in #{c_time})"
+  end
+
+
+  def run_phylo
+
+    if @software == "raxml"
+      if @aln_opt == "both"
+        raxml_tree_dna @bootstrap
+        raxml_tree_pep @bootstrap
+      elsif @aln_opt == "prot"
+        raxml_tree_pep @bootstrap
+      elsif @aln_opt == "dna"
+        raxml_tree_dna @bootstrap
+      end
+    elsif @software == "fasttree"
+      if @aln_opt == "both"
+        fasttree_tree_dna @bootstrap
+        fasttree_tree_pep @bootstrap
+      elsif @aln_opt == "prot"
+        fasttree_tree_pep @bootstrap
+      elsif @aln_opt == "dna"
+        fasttree_tree_dna @bootstrap
+      end
     end
 
   end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: bacterial-annotator
 version: !ruby/object:Gem::Version
-  version: 0.7.1
+  version: 0.8.0
 platform: ruby
 authors:
 - Maxime Deraspe