big_simon 0.1.0 → 0.1.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: dfbea3a58014cbec45a3959ad076fd842c6b393c
-  data.tar.gz: 69d9f8ac1dea196f64dd13fb06a13a4c26b38e45
+  metadata.gz: 9c81d1057b304f170cdf1dd6a7551bd981efc232
+  data.tar.gz: c0550efe308bc0080c9f14ce65057b100ff2f32b
 SHA512:
-  metadata.gz: 3a54fe903bb5c0f2f574a389dfc945ffe3d93bd2b1a15361bda81430542c2ef3990ac31e67b671690e056e23544e4f30f21dbecd756050dfb98dfb924fd15ca3
-  data.tar.gz: b61f6ba10b7efc267419ccccaef82654662923629e0b2b61e5a3c5e74b79fe8b753dd160c3e3d7fd57b1e926b54684f9367a839075544c496ff3a839ece57fb5
+  metadata.gz: 652abc1ab4507ed51012cde1a541a8e5ce75e30a2c2d8b9ef4cb0363e49bb8ea4f088dcb4786bb5f9f466f8f0ed0d2f2e6d9b446aa3a4c8fd2a6a8f4370f6a78
+  data.tar.gz: 3a82d66018a1dea01afb9f0230a9f8ad6ac90d96d5862a79b57ecb03f954eedf9183c2dc9d03b6f98a6a0675b3293577c4e5001fa7f792bbe68ebefe397d47f8

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    big_simon (0.1.0)
+    big_simon (0.1.1)
       parse_fasta (~> 2.5, >= 2.5.2)
       rya (~> 0.4.0)
       trollop (~> 2.1, >= 2.1.3)

data/Makefile

@@ -1,8 +1,8 @@
 test_small:
-	rm -r 0000TEST/; exe/big_simon -v spec/test_files/virus/* -h spec/test_files/host/* -o 0000TEST && tree 0000TEST
+	rm -r 0000TEST/; time exe/big_simon -t 3 -v spec/test_files/virus/* -h spec/test_files/host/* -o 0000TEST && tree 0000TEST
 
 test_small_install:
-	rm -r 0000TEST/; rake install && exe/big_simon -v spec/test_files/virus/* -h spec/test_files/host/* -o 0000TEST && tree 0000TEST
+	rm -r 0000TEST/; rake install && time exe/big_simon -t 3 -v spec/test_files/virus/* -h spec/test_files/host/* -o 0000TEST && tree 0000TEST
 
 test_toy:
 	rm -r toyexample_out; time exe/big_simon -v vendor/repos/VirHostMatcher/test/toyexample/virus/* -h vendor/repos/VirHostMatcher/test/toyexample/host/* -o toyexample_out -t 3

data/exe/big_simon

@@ -71,7 +71,7 @@ tmpdir_host  = File.join tmpdir, "host"
 # all_predictions_fname = File.join outdir, "scores_all.txt"
 mean_scaled_scores_fname = File.join outdir, "scores_scaled.mean.txt"
 
-virus_recs, host_recs = [], []
+# virus_recs, host_recs = [], []
 
 # Tempfile.open do |vir_f|
 #   Tempfile.open do |host_f|
@@ -159,38 +159,18 @@ name_map_host, all_ids_host   = BigSimon::Utils.set_up_tmp_dirs host_fnames, tmp
 wish_outf = BigSimon::Runners.wish BigSimon::WISH, tmpdir_virus, tmpdir_host, tmpdir, threads
 vhm_outf  = BigSimon::Runners.vir_host_matcher BigSimon::VHM, tmpdir_virus, tmpdir_host, tmpdir
 
+# TODO separate the parser from the runner for mummer.
 host_info_mummer = BigSimon::Runners.mummer BigSimon::MUMMER, tmpdir_virus, tmpdir_host, tmpdir, threads
 
-# Map them back to simple names.  TODO just have it spit these out from the beginning.
-host_info_mummer_simple_names = {}
-inverted_name_map_virus = name_map_virus.invert
-inverted_name_map_host = name_map_host.invert
-
-host_info_mummer.each do |virus, host_tables|
-  virname = virus
-  if inverted_name_map_virus.has_key? virus
-    virname = inverted_name_map_virus[virus]
-  end
-
-  host_info_mummer_simple_names[virname] = []
-
-  host_tables.map do |table|
-    hostname = inverted_name_map_host.has_key?(table[:host]) ? inverted_name_map_host[table[:host]] : table[:host]
-    new_table = { host: hostname, score: table[:score], scaled_score: table[:scaled_score] }
-
-    host_info_mummer_simple_names[virname] << new_table
-  end
-end
-
 host_info_wish = BigSimon::Parsers.wish wish_outf
 host_info_vhm  = BigSimon::Parsers.vir_host_matcher vhm_outf
 
-host_info_simple_names = BigSimon::Pipeline.collate_host_results [host_info_wish, host_info_vhm, host_info_mummer_simple_names], programs
+host_info_simple_names = BigSimon::Pipeline.collate_host_results [host_info_wish, host_info_vhm, host_info_mummer], programs
 host_info              = BigSimon::Pipeline.map_taxa host_info_simple_names, name_map_virus, name_map_host
 
-# puts
-# pp host_info
-# puts
+puts
+pp host_info
+puts
 
 # Just a basic all info file
 # File.open all_predictions_fname, "w" do |f|

data/exe/{ranks → consensus_predictions}

@@ -3,10 +3,15 @@ Signal.trap("PIPE", "EXIT")
 
 require "pp"
 
+require "big_simon"
+
+Rya::AbortIf.abort_unless ARGV.count >= 1, "usage: TOP=number consensus_predictions scores_scaled.*.txt > consensus_predictions.txt"
+
 by_program = {}
 lines = []
 
-TOP = (ENV["TOP"] || 5).to_i
+TOP = (ENV["TOP"] || 3).to_i
+AT_LEAST = (ENV["AT_LEAST"] || 2).to_i
 
 ARGV.each do |fname|
   scores = {}
@@ -78,16 +83,18 @@ by_program.each do |virus, program_tables|
   end
 end
 
-first_table.each do |virus, host_counts|
-  host_counts.sort_by { |_, programs| programs.count }.reverse.each do |host, programs|
-    STDERR.puts [virus, :best, host, programs.count, programs].join "\t"
-  end
-  STDERR.puts
-end
+# first_table.each do |virus, host_counts|
+#   host_counts.sort_by { |_, programs| programs.count }.reverse.each do |host, programs|
+#     STDERR.puts [virus, :best, host, programs.count, programs].join "\t"
+#   end
+#   STDERR.puts
+# end
 
 top_host_table.each do |virus, host_counts|
+  puts "\n\n>Virus-------#{virus}"
   host_counts.sort_by { |_, programs| programs.count }.reverse.each do |host, programs|
+    if programs.count >= AT_LEAST
       puts [virus, :top_N, host, programs.count, programs].join "\t"
+    end
   end
-  puts
 end

data/lib/big_simon/runners.rb

@@ -1,11 +1,118 @@
 require "tempfile"
+require "parse_fasta"
 
 module BigSimon
   class Runners
 
+    # @note To match the other things, you'd like them to be key'd on the file name.
+    def self.mummer exe, vir_dir, host_dir, outdir, threads
+      klass = Class.new.extend Rya::CoreExtensions::Math
+      FileUtils.mkdir_p outdir
+
+      mummer_outfname = File.join outdir, "mummer_out.txt"
+
+      virus_fnames = Dir.glob(vir_dir + "/*")
+      host_fnames  = Dir.glob(host_dir + "/*")
+
+      hit_table = {}
+
+      Tempfile.open do |vir_f|
+        Tempfile.open do |host_f|
+          virus_fnames.each do |fname|
+            Rya::AbortIf.assert fname.match(/.fa$/), "bad fname: #{fname}"
+
+            Object::ParseFasta::SeqFile.open(fname).each_record do |rec|
+              # id needs to be the file name
+              new_id = File.basename fname.sub(/.fa$/, "")
+
+              hit_table[new_id] = {}
+
+              vir_f.puts ">#{new_id}\n#{rec.seq}"
+
+              vir_f.puts ">#{new_id}___reverse\n#{rec.seq.reverse}"
+            end
+          end
+
+          host_fnames.each do |fname|
+            Rya::AbortIf.assert fname.match(/.fa$/), "bad fname: #{fname}"
+
+            Object::ParseFasta::SeqFile.open(fname).each_record do |rec|
+              new_id = File.basename fname.sub(/.fa$/, "")
+
+              # Add this host to each virus in the hit_table
+              hit_table.each do |virus, host_table|
+                host_table[new_id] = 0 # set it to defualt score of 0
+              end
+
+              host_f.puts ">#{new_id}\n#{rec.seq}"
+              host_f.puts ">#{new_id}___reverse\n#{rec.seq.reverse}"
+            end
+          end
+
+          vir_f.fsync
+          host_f.fsync
+
+          cmd = "mummer -threads #{threads} -qthreads #{threads} -maxmatch -l 15 #{host_f.path} #{vir_f.path} > #{mummer_outfname}"
+          Process.run_and_time_it! "MUMMER", cmd
+        end
+      end
+
+      virus = nil
+      overall_max_score = 0
+      File.open(mummer_outfname, "rt").each_line.with_index do |line, idx|
+        line.chomp!
+
+        unless line.empty?
+          if line.start_with? ">"
+            virus = line.chomp.sub(/^>/, "").sub(/___reverse$/, "").strip
+
+            # It can be duplicated as there are forward and reverse for each sequence (in case they're contigs.)
+
+            Rya::AbortIf.assert hit_table.has_key?(virus)
+            # unless hit_table.has_key? virus
+            #   hit_table[virus] = {}
+            # end
+          else
+            ary = line.strip.split " "
+
+            host = ary[0].sub(/___reverse$/, "").strip
+            score = ary[3].to_i
+
+            Rya::AbortIf.assert hit_table[virus].has_key?(host)
+
+            # unless hit_table[virus].has_key? host
+            #   hit_table[virus][host] = -1
+            # end
+
+            # We only want the longest hit.
+            hit_table[virus][host] = score if score > hit_table[virus][host]
+
+            # Track the overall max for scaling.
+            overall_max_score = score if score > overall_max_score
+          end
+        end
+      end
+
+      results_table = {}
+
+      min, max, from, to = 0, overall_max_score, 1, 0
+
+      hit_table.each do |virus, host_table|
+        results_table[virus] = []
+
+        host_table.each do |host, score|
+          scaled_score = klass.scale score, min, max, from, to
+
+          results_table[virus] << { host: host, score: score, scaled_score: scaled_score }
+        end
+      end
+
+      results_table
+    end
+
     # This one's a bit different as it parses as well and returns original names.
     # @todo Also do the reverse of each genome in case it's a contig.
-    def self.mummer exe, vir_dir, host_dir, outdir, threads
+    def self.mummer2 exe, vir_dir, host_dir, outdir, threads
       klass = Class.new.extend Rya::CoreExtensions::Math
       FileUtils.mkdir_p outdir
 

data/lib/big_simon/version.rb

@@ -1,5 +1,5 @@
 module BigSimon
-  VERSION = "0.1.0"
+  VERSION = "0.1.1"
 
   COPYRIGHT = "2018 Ryan Moore"
   CONTACT   = "moorer@udel.edu"

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: big_simon
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.1.1
 platform: ruby
 authors:
 - Ryan Moore
@@ -111,7 +111,7 @@ email:
 - moorer@udel.edu
 executables:
 - big_simon
-- ranks
+- consensus_predictions
 extensions: []
 extra_rdoc_files: []
 files:
@@ -129,7 +129,7 @@ files:
 - bin/console
 - bin/setup
 - exe/big_simon
-- exe/ranks
+- exe/consensus_predictions
 - lib/big_simon.rb
 - lib/big_simon/parsers.rb
 - lib/big_simon/pipeline.rb