lib/scaffolder/annotation_locator.rb in scaffolder-annotation-locator-0.0.1 vs lib/scaffolder/annotation_locator.rb in scaffolder-annotation-locator-0.1.0

@@ -1,48 +1,55 @@
 require 'delegate'
 require 'scaffolder'
 require 'bio'
 
+require 'scaffolder/extensions'
+
 class Scaffolder::AnnotationLocator < DelegateClass(Array)
 
   def initialize(scaffold_file,sequence_file,gff_file)
     @scaffold_file = scaffold_file
     @sequence_file = sequence_file
     @gff_file      = gff_file
 
     updated_records = Array.new
-    scaffold.inject(0) do |length,entry|
+    scaffold.inject(0) do |prior_length,entry|
 
       if entry.entry_type == :sequence
-        updated_records << records[entry.source].map do |record|
-          update_record(record,entry,length)
-        end
-      end
+        records[entry.source].each do |record|
 
-      length + entry.sequence.length
-    end
+          # Don't include this record if it overlaps with an insert
+          next if record.overlap?(entry.inserts.map{|i| (i.open..i.close)})
 
-    super updated_records.flatten
-  end
+          # Skip this record it lies in the start or stop trimmed regions
+          next if record.start < entry.start
+          next if record.end   > entry.stop
 
-  def update_record(record,scaffold_entry,prior_length)
-    record.start -= scaffold_entry.start - 1
-    record.end   -= scaffold_entry.start - 1
+          # Update record location by size differences of prior inserts
+          entry.inserts.select {|i| i.close < record.start }.each do |insert|
+            record.change_position_by insert.size_diff
+          end
 
-    if scaffold_entry.reverse
-      record.end   = scaffold_entry.sequence.length - (record.end - 1)
-      record.start = scaffold_entry.sequence.length - (record.start - 1)
+          # Decrease record position by distance contig is trimmed at start
+          record.change_position_by(1 - entry.start)
 
-      record.end, record.start = record.start, record.end
-      record.strand = self.class.flip_strand(record.strand)
-    end
+          # Reverse complement record positions if contig is reversed
+          record.reverse_complement_by entry.sequence.length if entry.reverse
 
-    record.start += prior_length
-    record.end   += prior_length
+          # Increase record position by length of prior contigs
+          record.change_position_by prior_length
 
-    record.seqname = "scaffold"
-    record
+          record.seqname = "scaffold"
+
+          updated_records << record
+        end
+      end
+
+      prior_length + entry.sequence.length
+    end
+
+    super updated_records
   end
 
   def scaffold
     Scaffolder.new(YAML.load(File.read(@scaffold_file)),@sequence_file)
   end
@@ -51,12 +58,8 @@
     gff3 = Bio::GFF::GFF3.new(File.read(@gff_file)).records
     gff3.inject(Hash.new{|h,k| h[k] = Array.new }) do |hash,record|
       hash[record.seqname] << record
       hash
     end
-  end
-
-  def self.flip_strand(strand)
-    strand == '+' ? '-' : '+'
   end
 
 end