crb-blast 0.6.6 → 0.6.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 275e3d2996f91648e9f046cd7a08c4c663243765
-  data.tar.gz: 4de455ec18f7c657585326d94e43dc761403cf77
+  metadata.gz: 0667a895d64d87a7ec7078342b9449d185d5fc8c
+  data.tar.gz: a0b4d4ffc83573e180beb7c84e778d21c3250155
 SHA512:
-  metadata.gz: b2725a63477308471676fbdabfd035fc83a469adede3ce56d49bd9020bf779ad53f6f3edec65e6ea744666dbe1df879116248f6ecfa8bdfa86dfdd4b393c2ca4
-  data.tar.gz: 6a731db6a14a8b8c24394891fe3335082e5b664e44314d18324f1072e76a78f37ec3152e459dc0500b293e836958bb84a6e9733961a77fb87a4cda16ee043d56
+  metadata.gz: c08e2203df9f0a077a14373b3e7f6a7b1b8d5281a7e55754448befe31af46f1247d66a7251c402610214f029f95ad02ae62074b431a9aefbc03d62b4eb8a5dd9
+  data.tar.gz: 8233a2eab64a5fceb841d601904b570eb160de41a6ddeeba5ea80a669823d3da9b15d9e0b095e37d6b47e09ae43cd1261028d33f4037d56dea6c9f91fdcc37e1

data/README.md

@@ -82,6 +82,23 @@ blaster.find_reciprocals
 blaster.find_secondaries
 ```
 
+### Output
+
+The output file for CRB-Blast pulls columns from the blast output.
+
+```
+query - the name of the transcript from the 'query' fasta file
+target - the name of the transcript from the 'target' fasta file
+id - the percent sequence identity
+alnlen - the alignment length
+evalue - the blast evalue
+bitscore - the blast bitscore
+qstart..qend - the coordinates of the alignment in the query from start to end
+tstart..tend - the coordinates of the alignment in the target from start to end 
+qlen - the length of the query transcript
+tlen - the length of the target transcript
+```
+
 ### Getting help
 
 Please use the issue tracker if you find bugs or have trouble running CRB-BLAST.

data/bin/crb-blast

@@ -50,6 +50,9 @@ EOS
   opt :split,
       "split the fasta files into chunks and run multiple blast jobs and then"+
       " combine them."
+
+  opt :verbose,
+      "be verbose"
 end
 
 Trollop::die :query, "must exist" if !File.exist?(opts[:query])
@@ -60,12 +63,27 @@ gem_deps = File.join(gem_dir, 'deps', 'deps.yaml')
 Bindeps.require gem_deps
 
 blaster = CRB_Blast::CRB_Blast.new(opts.query, opts.target)
+print "Making blast databases..." if opts.verbose
 dbs = blaster.makedb
+puts " Done" if opts.verbose
+
+print "Blasting..." if opts.verbose
 run = blaster.run_blast(opts.evalue, opts.threads, opts.split)
+puts " Done" if opts.verbose
+
+print "Loading..." if opts.verbose
 load = blaster.load_outputs
+puts " Done" if opts.verbose
+
+print "Finding reciprocals..." if opts.verbose
 recips = blaster.find_reciprocals
+puts " Done" if opts.verbose
+
+print "Fitting curve..." if opts.verbose
 secondaries = blaster.find_secondaries
+puts " Done" if opts.verbose
 
+print "Writing output..." if opts.verbose
 File.open("#{opts.output}", 'w') do |out|
   blaster.reciprocals.each_pair do |query_id, hits|
     hits.each do |hit|
@@ -73,3 +91,4 @@ File.open("#{opts.output}", 'w') do |out|
     end
   end
 end
+puts " Done" if opts.verbose

data/lib/crb-blast/crb-blast.rb

@@ -24,6 +24,7 @@ module CRB_Blast
     attr_accessor :missed
     attr_accessor :target_is_prot, :query_is_prot
     attr_accessor :query_results, :target_results, :working_dir
+    attr_reader :reciprocal_hits
 
     def initialize query, target, output=nil
       raise IOError.new("File not found #{query}") if !File.exist?(query)
@@ -189,12 +190,22 @@ module CRB_Blast
       cmd1 << " -out #{@output1} -evalue #{evalue} "
       cmd1 << " -outfmt \"6 std qlen slen\" "
       cmd1 << " -max_target_seqs 50 "
+      if bin1=="blastn"
+        cmd1 << " -dust no "
+      elsif bin1=~/blastx/ or bin1=~/blastp/ or bin1=~/tblastn/
+        cmd1 << " -seg no "
+      end
       cmd1 << " -num_threads #{threads}"
 
       cmd2 = "#{bin2} -query #{@target} -db #{@working_dir}/#{@query_name} "
       cmd2 << " -out #{@output2} -evalue #{evalue} "
       cmd2 << " -outfmt \"6 std qlen slen\" "
       cmd2 << " -max_target_seqs 50 "
+      if bin2=="blastn"
+        cmd2 << " -dust no "
+      elsif bin2=~/blastx/ or bin2=~/blastp/ or bin2=~/tblastn/
+        cmd2 << " -seg no "
+      end
       cmd2 << " -num_threads #{threads}"
       if !File.exist?("#{@output1}")
         blast1 = Cmd.new(cmd1)
@@ -432,9 +443,12 @@ module CRB_Blast
       else
         length_hash = Hash.new
         fitting = Hash.new
-        @evalues.each do |h|
-          length_hash[h[:length]] = [] if !length_hash.key?(h[:length])
-          length_hash[h[:length]] << h
+        File.open("evalues_data", "w") do |io|
+          @evalues.each do |h|
+            length_hash[h[:length]] = [] if !length_hash.key?(h[:length])
+            length_hash[h[:length]] << h
+            io.write "#{h[:length]}\t#{h[:e]}\n"
+          end
         end
 
         (10..@longest).each do |centre|
@@ -453,9 +467,24 @@ module CRB_Blast
           end
           if count>0
             mean = e/count
-            fitting[centre] = mean
+            if fitting[centre-1]
+              if fitting[centre-1] > mean # monotonic fitting
+                fitting[centre] = fitting[centre-1]
+              else
+                fitting[centre] = mean
+              end
+            else
+              fitting[centre] = mean
+            end
+          end
+        end
+        # output fitting data
+        File.open("fitting_data", "w") do |io|
+          fitting.each do |centre, mean|
+            io.write "#{centre}\t#{mean}\n"
           end
         end
+        #
         hits = 0
         @missed.each_pair do |id, list|
           list.each do |hit|
@@ -531,6 +560,8 @@ module CRB_Blast
     def write_output
       s=""
       unless @reciprocals.nil?
+      s << "query\ttarget\tid\talnlen\tevalue\tbitscore\t"
+      s << "qstart..qend\ttstart..tend\tqlen\ttlen\n"
         @reciprocals.each_pair do |query_id, hits|
           hits.each do |hit|
             s << "#{hit}\n"

data/lib/crb-blast/version.rb

@@ -3,7 +3,7 @@ module CRB_Blast
   module VERSION
     MAJOR = 0
     MINOR = 6
-    PATCH = 6
+    PATCH = 8
     BUILD = nil
 
     STRING = [MAJOR, MINOR, PATCH, BUILD].compact.join('.')

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: crb-blast
 version: !ruby/object:Gem::Version
-  version: 0.6.6
+  version: 0.6.8
 platform: ruby
 authors:
 - Chris Boursnell