gfa 0.6.1 → 0.6.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: cf18e4a80f49fc9c0ae4574c5c7bbc3c97fd41eaf5c7aff9fdd7991b7d562ccf
-  data.tar.gz: ac6999ffbbbeba45f9b81c6081ac16339ce4bb55502a77b116c77f91870af991
+  metadata.gz: 8069525ab289a4299c4d51df9a58a083df95292cd7bc59b83516ac4c5e7508f2
+  data.tar.gz: c1398a889d1f49431bfa01b196084ee9c448c60426fb179801151e7731538c03
 SHA512:
-  metadata.gz: c8ae1b273166574d8520aa1f747ac0dbd08e8f31846754ca6c3fa7f0432d5f31583aa915fcebb04a220e90fe14aeeb4f06dec2a5c23527571c9bb337de0ef9a5
-  data.tar.gz: f5fce0959cace1fe961d92fab2946bf8b67571ee6ceef3cd6e7047d8c46f0855b28e6b703ffbfe327ef8d09e1174b94ee98423ae36bb858bab76615849e0bd94
+  metadata.gz: d97f574a6081727f5687e259a954881351bf7767ec36d1f5ac8b2e6eef939a59a9406077619b306d9782e99357ec2cb98536b4e73474a9aad75c61614c7f7d13
+  data.tar.gz: 385a463ad705401ec083375b917d3654633f44e3828fcc93546422fb80712fc4bc22dbc2dd128dbdf1db44a8c60ed4d7a31df558db8786eb1fca12225a29b2ab

data/bin/gfa-paths-to-fasta

@@ -26,4 +26,11 @@ end
 $stderr.puts "Loading GFA: #{input}"
 gfa = GFA.load_parallel(input, (threads || 1).to_i)
 
+$stderr.puts "Saving path sequences: #{output}"
+File.open(output, 'w') do |fasta|
+  gfa.paths.set.each do |path|
+    fasta.puts '>%s' % path.path_name.value
+    fasta.puts path.sequence(gfa)
+  end
+end
 

data/lib/gfa/record/path.rb

@@ -42,4 +42,59 @@ class GFA::Record::Path < GFA::Record
 
     segment_names_a.any? { |name| segment.name == name }
   end
+
+  ##
+  # Array of GFA::Field::String with the sequences from each segment featuring
+  # the correct orientation from a +gfa+ (which *must* be indexed)
+  #
+  # TODO: Distinguish between a direct path (separated by comma) and a
+  # jump (separated by semicolon). Jumps include a distance estimate
+  # (column 6, optional) which could be used to add Ns between segment
+  # sequences (from GFA 1.2)
+  def segment_sequences(gfa)
+    raise "Unindexed GFA" unless gfa.indexed?
+    segment_names.value.split(/[,;]/).map do |i|
+      orientation = i[-1]
+      i[-1] = ''
+      segment = gfa.segments[i]
+
+      case orientation
+        when '+' ; segment.sequence
+        when '-' ; segment.rc
+        else ; raise "Unknown orientation: #{orientation} (path: #{path_name})"
+      end
+    end
+  end
+
+  ##
+  # Produce the contiguous path sequence based on the segment sequences and
+  # orientations from a +gfa+ (which *must* be indexed)
+  #
+  # TODO: Estimate gaps (Ns) from Jump distances (see +segment_sequences+)
+  #
+  # TODO: Attempt reading CIGAR values from the path first, the corresponding
+  # links next, and actually performing the pairwise overlap as last resort
+  #
+  # TODO: Support ambiguous IUPAC codes for overlap evaluation
+  def sequence(gfa)
+    segment_sequences(gfa).map(&:value)
+      .inject('') { |a, b| a + after_overlap(a, b) }
+  end
+
+  private
+    ##
+    # Find the overlap between sequences +a+ and +b+ (Strings) and return
+    # only the part of +b+ after the overlap. Assumes that +a+ starts
+    # at the same point or before +b+. If no overlap is found, returns +b+
+    # in its entirety.
+    def after_overlap(a, b)
+      (0 .. a.length - 1).each do |a_from|
+        a_to = b.length + a_from > a.length ? a.length : b.length + a_from
+        b_to = b.length + a_from > a.length ? a.length - a_from : b.length
+        if a[a_from .. a_to - 1] == b[0 .. b_to - 1]
+          return b[b_to .. b.length].to_s
+        end
+      end
+      b
+    end
 end

data/lib/gfa/record/segment.rb

@@ -32,4 +32,12 @@ class GFA::Record::Segment < GFA::Record
   def length
     sequence.value.length
   end
+
+  ##
+  # Returns the reverse-complement of the sequence (as a Z field)
+  def rc
+    GFA::Field::String.new(
+      sequence.value.upcase.reverse.tr('ACGTURYSWKMBDHVN', 'TGCAAYRSWMKVHDBN')
+    )
+  end
 end

data/lib/gfa/version.rb

@@ -1,5 +1,5 @@
 class GFA
-  VERSION       = '0.6.1'
+  VERSION       = '0.6.2'
   VERSION_ARRAY = VERSION.split(/\./).map { |x| x.to_i } # :nodoc:
   VERSION_MAJOR = VERSION_ARRAY[0] # :nodoc:
   VERSION_MINOR = VERSION_ARRAY[1] # :nodoc:

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: gfa
 version: !ruby/object:Gem::Version
-  version: 0.6.1
+  version: 0.6.2
 platform: ruby
 authors:
 - Luis M. Rodriguez-R
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2023-10-05 00:00:00.000000000 Z
+date: 2023-10-06 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rgl