bacterial-annotator 0.3.3 → 0.3.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 6a97539de7b1dff0da1ac8edaef6e0e10949c74e
-  data.tar.gz: 09fa573de375d54339adf3a8800318a49c059ba5
+  metadata.gz: 76bdda1755e5721c74d1d2207ce007d9a3b55410
+  data.tar.gz: 00adfe5d0405d2b7193be6b1e9a54826e2096946
 SHA512:
-  metadata.gz: 608daa4501f95457e6d412fdc43d0672e74acd4ecf91186ea57da00614033e569688ca2d6a82f89f4266bc62da2ca20d7703a9ace0d04f131de57025c93a7685
-  data.tar.gz: 6ac5946ce49c23bfeb0d32dbc38c41516df960ab73af5f6a1a99689a08f7076ecffb2e361cf4f087d11dd1dcb74858ab35c84aaac687d9a8b28112eee765c05e
+  metadata.gz: 399f59cfe68cc7fe8abdfa822c04825ff5168dbb7a6ab4bd73dcf01d539ac47d3e6dd983fe2df0539a0f3096333d26778c1e3bd4bf96a6225a103f62377066d6
+  data.tar.gz: 7519118185f6e197cd8985e06c9fddded3a1d3c36ce00a6b9684b0b4257de00b27283043fe4167438089c0f3b60962749d4fa8378816e9fc1928b7e617e99b4d

data/bin/ba_raxml

@@ -0,0 +1,46 @@
+#!/usr/bin/env ruby
+# -*- coding: utf-8 -*-
+# author:  	maxime déraspe
+# email:	maxime@deraspe.net
+# review:  	
+# date:    	15-02-24
+# version: 	0.01
+# licence:  	
+
+require 'open-uri'
+
+ROOT_path = File.dirname(__FILE__)
+# raxml_url = https://github.com/stamatak/standard-RAxML/archive/master.zip
+# Install RAXML on the user system
+def installRaxml
+
+  begin
+    resp = open("https://github.com/stamatak/standard-RAxML/archive/master.zip")
+    open("#{ROOT_path}/master.zip", "wb") do |file|
+      file.write(resp.read)
+    end
+    Dir.chdir("#{ROOT_path}/")
+    `unzip master.zip`
+    Dir.chdir("#{ROOT_path}/standard-RAxML-master")
+    `make -f Makefile.PTHREADS.gcc`
+    `rm *.o`
+    `cp #{ROOT_path}/standard-RAxML-master/raxmlHPC-PTHREADS #{ROOT_path}/raxml.linux`
+    File.chmod(0755, "#{ROOT_path}/raxml.linux")
+  rescue
+    abort "Problem in stalling RAXML, aborting"
+  end
+
+end
+
+
+# Install prodigal if not already install
+if ! File.exists? "#{ROOT_path}/raxml.linux"
+
+  puts "Installing RAXML git master.."
+  puts "See https://github.com/stamatak/standard-RAxML"
+  puts "License GPLv3 : https://github.com/stamatak/standard-RAxML/blob/master/gpl-3.0.txt"
+  installRaxml
+  puts "RAXML successfully installed in #{ROOT_path}/standard-RAxML-master"
+  puts ""
+
+end

data/bin/bacterial-annotator

@@ -132,9 +132,11 @@ compare [OPTIONS]
 
   //Alignment (MAFFT)
     --align        [dna|prot|both] by default align only proteins
-    --concat       <nb of genes | all>  by default all
+    --concat       <nb of genes>  by default 0=all
 
-  //Phylo (RAXML)
+  //Phylogeny (RAXML)
+    --phylogeny    will build phylogenetic tree from the alignments files (pep or dna)
+    --bootstrap    <nb of bootstrap> by default 100
 
 OEM
 
@@ -153,6 +155,9 @@ def parseOptions_compare
   options[:proc] = 2
   options[:align] = "prot"
   options[:genomes_list] = []
+  options[:concat] = 0
+  options[:phylogeny] = 0
+  options[:bootstrap] = 100
 
   while x = ARGV.shift
 
@@ -167,6 +172,12 @@ def parseOptions_compare
       options[:proc] = ARGV.shift
     when "--align"
       options[:align] = ARGV.shift
+    when "--concat"
+      options[:concat] = ARGV.shift
+    when "--phylogeny"
+      options[:phylogeny] = 1
+    when "--bootstrap"
+      options[:bootstrap] = (ARGV.shift).to_i
     when "--help", "-h"
       usage_compare
       abort
@@ -190,6 +201,7 @@ if ARGV.size > 1
   system("ba_prodigal")
   system("ba_blat")
   system("ba_mafft")
+  system("ba_raxml")
 
   options = {}
   genomes_list = []
@@ -223,14 +235,9 @@ if ARGV.size > 1
     ARGV.shift
     options = parseOptions_compare
     bcomp = BacterialComparator.new(options, ROOT)
-    if options[:align].downcase == "both"
-      bcomp.mafft_align_all_pep
-      bcomp.mafft_align_all_dna
-    elsif options[:align].downcase == "prot"
-      bcomp.mafft_align_all_pep
-    elsif options[:align].downcase == "dna"
-      bcomp.mafft_align_all_dna
-    end
+    aln_opt = options[:align].downcase
+    bcomp.mafft_aln aln_opt
+    bcomp.raxml_tree aln_opt, options[:bootstrap] if options[:phylogeny] == 1
 
   else
 

data/lib/bacterial-comparator.rb

@@ -84,8 +84,8 @@ class BacterialComparator
 
   def build_multifasta ref_prot, synteny
 
-    pep_out_dir = "./#{@outdir}/genes-align-pep"
-    dna_out_dir = "./#{@outdir}/genes-align-dna"
+    pep_out_dir = "./#{@outdir}/align-genes-pep"
+    dna_out_dir = "./#{@outdir}/align-genes-dna"
 
     # create multifasta by syntenic proteins (pep)
     if ! File.exists? pep_out_dir+"/#{ref_prot}.pep"
@@ -160,8 +160,8 @@ class BacterialComparator
 
     fout.close
 
-    pep_out_dir = "./#{@outdir}/genes-align-pep"
-    dna_out_dir = "./#{@outdir}/genes-align-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
 
@@ -202,8 +202,10 @@ class BacterialComparator
 
 
   def mafft_align_all_pep
+
     puts "# MAFFT multialign all protein sequences.."
-    Dir.chdir("#{@outdir}/genes-align-pep/")
+    ori_dir = Dir.pwd
+    Dir.chdir("#{@outdir}/align-genes-pep/")
 
     is_done = 1
     if Dir["*.pep"].length == Dir["*.aln"].length
@@ -220,16 +222,19 @@ class BacterialComparator
       Parallel.map(Dir["*.pep"], in_processes: @proc) { |f|
         mafft_align f
       }
+    else
+      puts "..Prot alignment files already exists, skipping."
     end
 
-    concat_alignments "genes-align-pep.concat.fasta"
-    Dir.chdir("../../")
+    concat_alignments "align-genes-pep.all.fasta"
+    Dir.chdir(ori_dir)
 
   end
 
   def mafft_align_all_dna
     puts "# MAFFT multialign all gene sequences.."
-    Dir.chdir("#{@outdir}/genes-align-dna/")
+    ori_dir = Dir.pwd
+    Dir.chdir("#{@outdir}/align-genes-dna/")
 
     is_done = 1
     if Dir["*.dna"].length == Dir["*.aln"].length
@@ -246,16 +251,23 @@ class BacterialComparator
       Parallel.map(Dir["*.dna"], in_processes: @proc) { |f|
         mafft_align f
       }
+    else
+      puts "..Gene alignment files already exists, skipping."
     end
 
-    concat_alignments "genes-align-dna.concat.fasta"
-    Dir.chdir("../../")
+    concat_alignments "align-genes-dna.all.fasta"
+    Dir.chdir(ori_dir)
 
   end
 
 
   def concat_alignments outfile
 
+    if File.exists?("../#{outfile}") and File.size("../#{outfile}") > 0
+      puts "..Alignment concatenated file already exists, skipping."
+      return
+    end
+
     fout = File.open("../#{outfile}", "w")
 
     ref_id = Dir["../../#{@genomes_list[0]}/*.pep"][0].gsub(/.*\//,"").gsub(".pep","")
@@ -297,5 +309,64 @@ class BacterialComparator
 
   end
 
+  def mafft_aln aln_opt
+
+    if aln_opt == "both"
+      mafft_align_all_pep
+      mafft_align_all_dna
+    elsif aln_opt == "prot"
+      mafft_align_all_pep
+    elsif aln_opt == "dna"
+      mafft_align_all_dna
+    end
+
+  end
+
+
+  def raxml_tree_dna bt
+
+    # DNA tree
+    puts "# RAXML DNA tree creation.. "
+    ori_dir = Dir.pwd
+    Dir.chdir(@outdir)
+    Dir.mkdir("tree-genes-dna") if ! Dir.exists?("tree-genes-dna")
+    current_dir = Dir.pwd
+    tree_dir = "#{current_dir}/tree-genes-dna"
+    cmd = system("#{@root}/raxml.linux -T #{@proc} -f d -N #{bt} -s align-genes-dna.all.fasta  -m GTRGAMMA -p 123454321 -n DnaTree -w #{tree_dir}")
+    cmd = system("cat #{tree_dir}/RAxML_result.DnaTree.RUN.* >> #{tree_dir}/RAxML_result.BS")
+    cmd = system("#{@root}/raxml.linux -T 3 -f b -z #{tree_dir}/RAxML_result.BS -t #{tree_dir}/RAxML_bestTree.DnaTree -m GTRGAMMA -n DNA_BS_TREE -w #{tree_dir}")
+    cmd = system("ln -s #{tree_dir}/RAxML_bipartitionsBranchLabels.DNA_BS_TREE #{tree_dir}/../")
+    Dir.chdir(ori_dir)
+  end
+
+  def raxml_tree_pep bt
+
+    # Prot tree
+    puts "# RAXML Protein tree creation.. "
+    ori_dir = Dir.pwd
+    Dir.chdir(@outdir)
+    Dir.mkdir("tree-genes-pep") if ! Dir.exists?("tree-genes-pep")
+    current_dir = Dir.pwd
+    tree_dir = "#{current_dir}/tree-genes-pep"
+    cmd = system("#{@root}/raxml.linux -T #{@proc} -f d -N #{bt} -s align-genes-pep.all.fasta  -m PROTGAMMAAUTO -p 123454321 -n PepTree -w #{tree_dir}")
+    cmd = system("cat #{tree_dir}/RAxML_result.PepTree.RUN.* >> #{tree_dir}/RAxML_result.BS")
+    cmd = system("#{@root}/raxml.linux -T 3 -f b -z #{tree_dir}/RAxML_result.BS -t #{tree_dir}/RAxML_bestTree.PepTree -m PROTGAMMAAUTO -n PEP_BS_TREE -w #{tree_dir}")
+    cmd = system("ln -s #{tree_dir}/RAxML_bipartitionsBranchLabels.PEP_BS_TREE #{tree_dir}/../")
+    Dir.chdir(ori_dir)
+  end
+
+
+  def raxml_tree aln_opt, bt
+
+    if aln_opt == "both"
+      raxml_tree_dna bt
+      raxml_tree_pep bt
+    elsif aln_opt == "prot"
+      raxml_tree_pep bt
+    elsif aln_opt == "dna"
+      raxml_tree_dna bt
+    end
+
+  end
 
 end                             # end of Class

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: bacterial-annotator
 version: !ruby/object:Gem::Version
-  version: 0.3.3
+  version: 0.3.4
 platform: ruby
 authors:
 - Maxime Deraspe
@@ -78,12 +78,14 @@ executables:
 - ba_prodigal
 - ba_blat
 - ba_mafft
+- ba_raxml
 extensions: []
 extra_rdoc_files: []
 files:
 - bin/ba_blat
 - bin/ba_mafft
 - bin/ba_prodigal
+- bin/ba_raxml
 - bin/bacterial-annotator
 - lib/bacterial-annotator.rb
 - lib/bacterial-annotator/fasta-manip.rb