in progress

diff --git a/forester/ruby/evoruby/lib/evo/apps/domain_sequence_extractor_new.rb b/forester/ruby/evoruby/lib/evo/apps/domain_sequence_extractor_new.rb
index efa3968..098cf05 100644 (file)
--- a/forester/ruby/evoruby/lib/evo/apps/domain_sequence_extractor_new.rb
+++ b/forester/ruby/evoruby/lib/evo/apps/domain_sequence_extractor_new.rb
@@ -30,6 +30,7 @@ module Evoruby
     ADD_DOMAIN_NUMBER_OPTION           = 'd'
     ADD_DOMAIN_NUMBER_OPTION_AS_DIGIT  = 'dd'
     ADD_DOMAIN_NUMBER_OPTION_AS_LETTER = 'dl'
     ADD_DOMAIN_NUMBER_OPTION           = 'd'
     ADD_DOMAIN_NUMBER_OPTION_AS_DIGIT  = 'dd'
     ADD_DOMAIN_NUMBER_OPTION_AS_LETTER = 'dl'

+    ADD_SPECIES                        = 's'

     TRIM_OPTION                        = 't'
     LOG_FILE_SUFFIX                    = '_domain_seq_extr.log'
     PASSED_SEQS_SUFFIX                 = '_domain_seq_extr_passed'
     TRIM_OPTION                        = 't'
     LOG_FILE_SUFFIX                    = '_domain_seq_extr.log'
     PASSED_SEQS_SUFFIX                 = '_domain_seq_extr_passed'

@@ -112,6 +113,11 @@ module Evoruby
         add_domain_number_as_digit = true
       end
 
         add_domain_number_as_digit = true
       end
 

+      add_species = false
+      if cla.is_option_set? ADD_SPECIES 
+        add_species = true
+      end
+      

       if ( add_domain_number_as_letter && add_domain_number_as_digit )
         puts( "attempt to add domain number as letter and digit at the same time" )
         print_help
       if ( add_domain_number_as_letter && add_domain_number_as_digit )
         puts( "attempt to add domain number as letter and digit at the same time" )
         print_help

@@ -217,6 +223,7 @@ module Evoruby
           add_domain_number_as_digit,
           add_domain_number_as_letter,
           trim,
           add_domain_number_as_digit,
           add_domain_number_as_letter,
           trim,

+           add_species,

           log )
       rescue ArgumentError, IOError, StandardError => e
         Util.fatal_error( PRG_NAME, "error: " + e.to_s, STDOUT )
           log )
       rescue ArgumentError, IOError, StandardError => e
         Util.fatal_error( PRG_NAME, "error: " + e.to_s, STDOUT )

@@ -258,6 +265,7 @@ module Evoruby
       puts( "           -" + ADD_DOMAIN_NUMBER_OPTION_AS_DIGIT  + ": to add numbers to extracted domains as digit (example \"domain2\")" )
       puts( "           -" + ADD_DOMAIN_NUMBER_OPTION_AS_LETTER  + ": to add numbers to extracted domains as letter (example \"domaina\")" )
       puts( "           -" + TRIM_OPTION  + ": to remove the last 2 characters from sequence names" )
       puts( "           -" + ADD_DOMAIN_NUMBER_OPTION_AS_DIGIT  + ": to add numbers to extracted domains as digit (example \"domain2\")" )
       puts( "           -" + ADD_DOMAIN_NUMBER_OPTION_AS_LETTER  + ": to add numbers to extracted domains as letter (example \"domaina\")" )
       puts( "           -" + TRIM_OPTION  + ": to remove the last 2 characters from sequence names" )

+      puts( "           -" + ADD_SPECIES  + ": to add species [in brackets]" )

       puts()
     end
 
       puts()
     end
 

diff --git a/forester/ruby/evoruby/lib/evo/io/parser/hmmscan_domain_extractor.rb b/forester/ruby/evoruby/lib/evo/io/parser/hmmscan_domain_extractor.rb
index bd3ae8b..3bf3fb7 100644 (file)
--- a/forester/ruby/evoruby/lib/evo/io/parser/hmmscan_domain_extractor.rb
+++ b/forester/ruby/evoruby/lib/evo/io/parser/hmmscan_domain_extractor.rb
@@ -37,6 +37,7 @@ module Evoruby
         add_domain_number_as_digit,
         add_domain_number_as_letter,
         trim_name,
         add_domain_number_as_digit,
         add_domain_number_as_letter,
         trim_name,

+         add_species,

         log )
 
       Util.check_file_for_readability( hmmsearch_output )
         log )
 
       Util.check_file_for_readability( hmmsearch_output )

@@ -225,7 +226,8 @@ module Evoruby
         add_domain_number,
         add_domain_number_as_digit,
         add_domain_number_as_letter,
         add_domain_number,
         add_domain_number_as_digit,
         add_domain_number_as_letter,

-        trim_name )
+        trim_name,
+        add_species      )

       if ( number < 1 || out_of < 1 || number > out_of )
         error_msg = "impossible: number=" + number.to_s + ", out of=" + out_of.to_s
         raise ArgumentError, error_msg
       if ( number < 1 || out_of < 1 || number > out_of )
         error_msg = "impossible: number=" + number.to_s + ", out of=" + out_of.to_s
         raise ArgumentError, error_msg

@@ -246,7 +248,9 @@ module Evoruby
       # hmmsearch is 1 based, wheres sequences are 0 bases in this package.
       seq = in_msa.get_sequence( seqs[ 0 ] ).get_subsequence( seq_from - 1, seq_to - 1 )
       
       # hmmsearch is 1 based, wheres sequences are 0 bases in this package.
       seq = in_msa.get_sequence( seqs[ 0 ] ).get_subsequence( seq_from - 1, seq_to - 1 )
       

-      seq.set_name( seq.get_name.split[ 0 ] )
+      orig_name = seq.get_name
+      
+      seq.set_name( orig_name.split[ 0 ] )

       
       if add_position
         seq.set_name( seq.get_name + "_" + seq_from.to_s + "-" + seq_to.to_s )
       
       if add_position
         seq.set_name( seq.get_name + "_" + seq_from.to_s + "-" + seq_to.to_s )

@@ -268,6 +272,18 @@ module Evoruby
         elsif ( add_domain_number )
           seq.set_name( seq.get_name + "~" + number.to_s + "-" + out_of.to_s )
         end
         elsif ( add_domain_number )
           seq.set_name( seq.get_name + "~" + number.to_s + "-" + out_of.to_s )
         end

+        
+        if add_species
+          a = orig_name.rindex "["
+          b = orig_name.rindex "]"
+          unless a && b
+               error_msg = "species not found in " + orig_name
+               raise StandardError, error_msg
+          end
+          species = orig_name[ a .. b ]
+          seq.set_name( seq.get_name + " [" + species + "]" )
+        end  
+          

       end
 
       # if ( seq.get_name.length > 10 )
       end
 
       # if ( seq.get_name.length > 10 )