gfa 0.9.2 → 0.9.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 80f578050737b697fa9205a6e66dd5a3206c3c8e49dad24f066e76444829334a
-  data.tar.gz: 57d8d6436d92e009c867ac1939e2b9522b392931f3c404d0f2964b58111bb0d1
+  metadata.gz: 8919a697174aabef8c92bf7024d855c31c2adce2402e05474882cc03754e9ce2
+  data.tar.gz: 3a03b9fd5446938df35974eb0e225a35ecdc69539546ff0c6873fdba0cb28928
 SHA512:
-  metadata.gz: 96145760f6f613ebfab88463c3dc2587dfd319ad7deccaec943b7ce93491b974ef5045e63b11ecf0c6cb4583b7b92c97b0132465838b7648e8bb1cd7e59cb5f6
-  data.tar.gz: bc1ae46b2da9278db465afff02995679da7b4c71a558cea04396a8343eb998f4e28cbc1cb729886f030cbb1fe6093edca73945d85307a1cbf2a3bfe3aed74cef
+  metadata.gz: f72d49eb2fa08a89a8bf919cbc73161f65a1c2d083aa783856f33a5d872895a450d6de87a52977bfe69e5de35b0677cd00473b73d6a515449407c3705cfa3744
+  data.tar.gz: b57b09ffd99e0744fbf686d328acf778d4dcb502ae88cb79cfc3722d2c0b0d415569db0a9cce27367e6c16048ac3788efd7d4af3bfa2f3bbce7e66fad8b0f776

data/bin/gfa-add-gaf

@@ -34,7 +34,8 @@ unless degree
 end
 
 $stderr.puts "Loading GFA: #{input_gfa}"
-gfa = GFA.load_parallel(input_gfa, (threads || 1).to_i)
+threads = (threads || 1).to_i
+gfa = GFA.load_parallel(input_gfa, threads)
 
 $stderr.puts "Loading GAF: #{input_gaf}"
 $stderr.puts "- Minimum identity: #{0.95}"
@@ -65,7 +66,7 @@ $stderr.puts "- Found #{segments.size} linked segments"
 degree = degree.to_i
 if degree >= 0
   $stderr.puts 'Subsetting graph'
-  gfa = gfa.subgraph(segments, degree: degree)
+  gfa = gfa.subgraph(segments, degree: degree, threads: threads)
 end
 
 $stderr.puts "Saving GFA: #{output}"

data/lib/gfa/common.rb

@@ -22,7 +22,6 @@ class GFA
 
   def self.advance
     @advance_bar_i += 1
-    # $stderr.print "#{@advance_bar_i}"[-1] + "\b"
     while 50 * @advance_bar_i / @advance_bar_n > @advance_bar_p
       $stderr.print "\b=>"
       @advance_bar_p += 1
@@ -33,17 +32,16 @@ class GFA
     t_t = Time.now - @advance_bar_s
     t_u = 'sec'
 
-    if t_t > 60
-      t_t /= 60
-      t_u = 'min'
+    [[60, 'min'], [60, 'h'], [24, 'd'], [30, 'mo']].each do |t_a|
+      break if t_t < t_a[0]
+      t_t /= t_a[0]
+      t_u  = t_a[1]
     end
 
-    if t_t > 60
-      t_t /= 60
-      t_u = 'h'
-    end
-
-    $stderr.puts '  [ %-48s ]' % "Time elapsed: #{'%.1f' % t_t} #{t_u}"
+    ram_gb = `ps -o rss= -p #{$$}`.to_f / 1024 / 1024
+    info1 = 'Time elapsed: %.1f %s' % [t_t, t_u]
+    info2 = '%.1fG RAM' % ram_gb
+    $stderr.puts '  [ %-36s%12s ]' % [info1, info2]
   end
 
   # Instance-level

data/lib/gfa/graph.rb

@@ -62,10 +62,13 @@ class GFA
   # If +headers+, it includes all the original headers. Otherwise it only
   # only includes the version header (might be inferred).
   #
+  # +threads+ indicates the number of threads to use in the processing of this
+  # operation, currently only affecting expansion rounds.
+  #
   # All comments are ignored even if originally parsed. Walks are currently
   # ignored too. If the current GFA object doesn't have an index, it builds one
   # and forces +index: true+. The output object inherits all options.
-  def subgraph(segments, degree: 1, headers: true)
+  def subgraph(segments, degree: 1, headers: true, threads: 2)
     # Prepare objects
     unless opts[:index]
       opts[:index] = true
@@ -89,7 +92,7 @@ class GFA
     segments.each { |segment| gfa << segment }
 
     # Expand graph
-    linking = linking_records(gfa.segments, degree: degree)
+    linking = linking_records(gfa.segments, degree: degree, threads: threads)
     linking.each { |record| gfa << record }
 
     # Return
@@ -99,23 +102,17 @@ class GFA
   ##
   # Finds all the records linking to any segments in +segments+, a
   # GFA::RecordSet::SegmentSet object, and expands to links with up to
-  # +degree+ degrees of separation
-  #
-  # It only evaluates the edges given in the +edges+ Array of GFA::Record
-  # values. If +edges+ is +nil+, it uses the full set of edges in the gfa.
-  # Edge GFA::Record objects can be of type Link, Containment, Jump, or Path
-  #
-  # If +_ignore+ is passed, it ignores this number of segments at the beginning
-  # of the +segments+ set (assumes they have already been evaluated). This is
-  # only used for internal heuristics
+  # +degree+ degrees of separation. This operation uses +threads+ Threads
+  # (with shared RAM)
   #
   # Returns an array of GFA::Record objects with all the identified linking
-  # records (edges)
+  # records (edges). Edge GFA::Record objects can be of type: Link, Containment,
+  # Jump, or Path
   #
-  # IMPORTANT NOTE 1: The object +segments+ will be modified to include all
+  # IMPORTANT NOTE: The object +segments+ will be modified to include all
   # linked segments. If you don't want this behaviour, please make sure to pass
-  # a duplicate of the object instead.
-  def linking_records(segments, degree: 1)
+  # a duplicate of the object instead
+  def linking_records(segments, degree: 1, threads: 2)
     unless segments.is_a? GFA::RecordSet::SegmentSet
       raise "Unrecognised class: #{segments.class}"
     end
@@ -123,17 +120,43 @@ class GFA
     edge_matrix unless degree == 0 # Just to trigger matrix calculation
     degree.times do |round|
       $stderr.puts "- Expansion round #{round + 1}"
-      self.class.advance_bar(segments.size)
+      self.class.advance_bar(segments.size + 1)
       pre_expansion = segments.size
-      new_segments = []
-      segments.set.each do |segment|
-        self.class.advance
-        idx = self.segments.position(segment)
-        edge_matrix[nil, idx].each_with_index do |edge, target_k|
-          new_segments << target_k if edge
+
+      # Launch children processes
+      io  = []
+      pid = []
+      threads.times do |t|
+        io[t] = IO.pipe
+        pid << fork do
+          new_segments = Set.new
+          segments.set.each_with_index do |segment, k|
+            self.class.advance if t == 0
+            next unless (k % threads) == t
+            idx = self.segments.position(segment)
+            edge_matrix[nil, idx].each_with_index do |edge, target_k|
+              new_segments << target_k if edge
+            end
+          end
+          Marshal.dump(new_segments, io[t][1])
+          self.class.advance if t == 0
+          exit!(0)
         end
+        io[t][1].close
+      end
+
+      # Collect and merge results
+      new_segments = Set.new
+      io.each_with_index do |pipe, k|
+        result = pipe[0].read
+        Process.wait(pid[k])
+        self.class.advance_bar(io.size) if k == 0
+        raise "Child process failed: #{k}" if result.empty?
+        new_segments += Marshal.load(result)
+        pipe[0].close
+        self.class.advance
       end
-      new_segments = new_segments.uniq.map { |i| self.segments[i] }
+      new_segments = new_segments.map { |i| self.segments[i] }
       new_segments.each { |i| segments << i unless segments[i.name] }
       new = segments.size - pre_expansion
       $stderr.puts "  #{new} segments found, total: #{segments.size}"

data/lib/gfa/parser.rb

@@ -39,7 +39,7 @@ class GFA
   # and the same +opts+ supported by +load+. Defaults to the +load+ method
   # instead if +thr <= 1+.
   def self.load_parallel(file, thr, opts = {})
-    return self.load(file, opts) if thr <= 1
+    return self.load(file, opts) if thr < 1
 
     # Prepare data
     lno = 0
@@ -48,7 +48,7 @@ class GFA
     blk = (lno.to_f / thr).ceil
 
     # Launch children processes
-    advance_bar(blk)
+    advance_bar(blk + 1)
     io  = []
     pid = []
     thr.times do |i|
@@ -58,10 +58,11 @@ class GFA
         o = opts.merge(line_range: [i * blk, (i + 1) * blk - 1])
         records = []
         read_records(file, o) do |record|
-          advance if i == 0
           records << record
+          advance if i == 0
         end
         Marshal.dump(records, io[i][1])
+        advance if i == 0
         exit!(0)
       end
       io[i][1].close
@@ -72,9 +73,11 @@ class GFA
     io.each_with_index do |pipe, k|
       result = pipe[0].read
       Process.wait(pid[k])
+      advance_bar(io.size) if k == 0
       raise "Child process failed: #{k}" if result.empty?
       Marshal.load(result).each { |record| gfa << record }
       pipe[0].close
+      advance
     end
 
     return gfa

data/lib/gfa/version.rb

@@ -1,5 +1,5 @@
 class GFA
-  VERSION       = '0.9.2'
+  VERSION       = '0.9.3'
   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: 0.9.2
+  version: 0.9.3
 platform: ruby
 authors:
 - Luis M. Rodriguez-R
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2024-07-03 00:00:00.000000000 Z
+date: 2024-08-30 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rgl