lib/npsearch/signalp.rb in npsearch-2.1.0 vs lib/npsearch/signalp.rb in npsearch-2.1.1

@@ -1,6 +1,8 @@
 require 'forwardable'
+require 'open3'
+require 'timeout'
 
 # Top level module / namespace.
 module NpSearch
   # A class to hold sequence data
   class Signalp
@@ -9,36 +11,37 @@
       def_delegators NpSearch, :opt
 
       def analyse_sequence(seq)
         sp_headers = %w(name cmax cmax_pos ymax ymax_pos smax smax_pos smean d
                         sp dmaxcut networks orf)
-        data       = setup_analysis(seq)
-        orf_results = []
-        s = `echo "#{data[:fasta]}\n" | #{opt[:signalp_path]} -t euk \
-             -f short -U 0.34 -u 0.34`
-        sp_results = s.split("\n").delete_if { |l| l[0] == '#' }
-        sp_results.each_with_index do |line, idx|
-          line = line + ' ' + data[:seq][idx].to_s
-          orf_results << Hash[sp_headers.map(&:to_sym).zip(line.split)]
+        seqs       = setup_analysis(seq)
+        sp_results = []
+        seqs.each do |seq|
+          sp_results << run_signalp(seq, sp_headers)
         end
-        orf_results.sort_by { |h| h[:d] }.reverse[0]
+        sp_results.sort_by { |h| h[:d] }.reverse[0]
       end
 
-      def setup_analysis(seq)
-        if opt[:type] == :protein
-          data = { seq: [seq], fasta: ">seq\n#{seq}" }
-        else
-          orfs = seq.scan(/(?=(M\w+))./).flatten
-          orfs.unshift(seq)
-          data = { seq: orfs, fasta: create_orf_fasta(orfs) }
+      private
+
+      def run_signalp(seq, sp_headers)
+        Timeout::timeout(300) do
+          cmd = "echo '>seq\n#{seq}\n' | #{opt[:signalp_path]} -t euk" \
+                " -f short -U 0.34 -u 0.34"
+          stdin, stdout, stderr, wait_thr = Open3.popen3(cmd)
+          out = stdout.gets(nil).split("\n").delete_if { |l| l[0] == '#' }
+          stdin.close; stdout.close; stderr.close
+          result = out[0] + ' ' + seq
+          return Hash[sp_headers.map(&:to_sym).zip(result.split)]
         end
-        data
+      rescue Timeout::Error
+        no_results = [0,0,1,1,1,1,1,1,1,'N',1,1, seq]
+        return Hash[sp_headers.map(&:to_sym).zip(no_results)]
       end
 
-      def create_orf_fasta(m_orf)
-        fasta = ''
-        m_orf.each_with_index { |seq, idx| fasta << ">#{idx}\n#{seq}\n" }
-        fasta
+      def setup_analysis(seq)
+        orfs = seq.scan(/(?=(M\w{#{opt[:min_orf_length]},}))./).flatten
+        (opt[:type] == :protein || orfs.empty? || orfs.nil?) ? [seq] : orfs
       end
     end
   end
 end