in progress...

[jalview.git] / forester / ruby / evoruby / lib / evo / msa / msa.rb

diff --git a/forester/ruby/evoruby/lib/evo/msa/msa.rb b/forester/ruby/evoruby/lib/evo/msa/msa.rb
index 0dcad71..1cdc78d 100644 (file)
--- a/forester/ruby/evoruby/lib/evo/msa/msa.rb
+++ b/forester/ruby/evoruby/lib/evo/msa/msa.rb
@@ -4,7 +4,7 @@
 # Copyright::    Copyright (C) 2017 Christian M. Zmasek
 # License::      GNU Lesser General Public License (LGPL)
 #
-# Last modified: 2017/02/07
+# Last modified: 2017/03/13
 
 require 'lib/evo/util/constants'
 require 'lib/evo/util/util'
@@ -17,6 +17,15 @@ module Evoruby
       @identical_seqs_detected = Array.new
       @name_to_seq_indices = Hash.new
       @namestart_to_seq_indices = Hash.new
+      @name = nil
+    end
+
+    def get_name
+      @name
+    end
+
+    def set_name( name )
+      @name = name
     end
 
     def add_sequence( sequence )
@@ -397,7 +406,7 @@ module Evoruby
       removed
     end
 
-    def trim!( first, last )
+    def trim!( first, last, name_suffix = nil )
       cols = Array.new()
       for i in 0 ... get_length()
         if ( i < first || i > last )
@@ -405,6 +414,58 @@ module Evoruby
         end
       end
       remove_columns!( cols )
+      if name_suffix != nil
+        n = get_number_of_seqs
+        for s in 0 ... n
+          seq = get_sequence( s )
+          seq.set_name( seq.get_name() + name_suffix );
+        end
+      end
+    end
+
+    def extract( first, last )
+      if !is_aligned()
+        error_msg = "attempt to extract from unaligned msa"
+        raise StandardError, error_msg, caller
+      end
+      if first < 0
+        error_msg = "first < 0"
+        raise StandardError, error_msg, caller
+      end
+      if last >= get_length()
+        error_msg = "last > length"
+        raise StandardError, error_msg, caller
+      end
+      if first >= last
+        error_msg = "first >= last"
+        raise StandardError, error_msg, caller
+      end
+      msa = Msa.new()
+      for i in 0 ... get_number_of_seqs
+        msa.add_sequence( get_sequence( i ).get_subsequence( first, last ) )
+      end
+      msa
+    end
+
+    def sliding_extraction( step, size )
+      counter = 0
+      done = false
+      msas = Array.new()
+      while !done
+        first = counter * step
+        last = first + size - 1
+        if last > get_length() - 1
+          last = get_length() - 1
+          done = true
+        end
+        unless first >= last
+          counter +=1
+          res = extract( first, last)
+          res.set_name(first.to_s + "-" + last.to_s)
+          msas << res
+        end
+      end
+      msas
     end
 
     def get_gap_only_columns()
@@ -494,7 +555,7 @@ module Evoruby
       return cols
     end
 
-    # Split an alignment into n alignemnts of equal size, except last one
+    # Split an alignment into n alignmnts of equal size, except last one
     def split( n, verbose = false )
       if ( n < 2 || n > get_number_of_seqs )
         error_msg = "attempt to split into less than two or more than the number of sequences"
@@ -527,6 +588,35 @@ module Evoruby
       msas
     end
 
+    def split_by_os(verbose = false)
+      msa_hash = Hash.new()
+      for i in 0 ... get_number_of_seqs()
+        seq = get_sequence(i)
+        name = seq.get_name()
+        # >sp|Q1HVE7|AN_EBVA8 Shutoff alkaline exonuclease OS=Epstein-Barr virus (strain AG876) GN=BGLF5 PE=3 SV=1
+        # if name =~ /OS=(.+?)\s+[A-Z]{2}=/
+        if name =~ /Organism:(.+?)(\|Protein|$)/
+          os = $1
+          unless  msa_hash.has_key?(os)
+            msa_hash[os] = Msa.new
+          end
+          msa_hash[os].add_sequence seq
+        else
+          error_msg = "sequence name \"" + name + "\" is not in the expected format for splitting by OS"
+          raise IOError, error_msg, caller
+        end
+      end
+      msa_hash = msa_hash.sort{|a, b|a<=>b}.to_h
+      if verbose
+        c = 0
+        msa_hash.each do |o, msa|
+          c += 1
+          puts c.to_s + ': ' + o
+        end
+      end
+      msa_hash
+    end
+
     private
 
     def get_overlaps( min_overlap = 1 )