exe/big_simon in big_simon-0.1.1 vs exe/big_simon in big_simon-0.2.0

@@ -52,11 +52,12 @@
                           "--threads must be > 0"
 
 programs = [
   "WIsH",
   "VirHostMatcher",
-  "mummer"
+  "mummer",
+  "homology",
 ]
 
 outdir       = opts[:outdir]
 threads      = opts[:threads]
 virus_fnames = opts[:viruses]
@@ -69,127 +70,96 @@
 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 = [], []
 
-# Tempfile.open do |vir_f|
-#   Tempfile.open do |host_f|
-#     virus_fnames.each do |fname|
-#       ParseFasta::SeqFile.open(fname).each_record do |rec|
-#         vir_f.puts rec
-#
-#         vir_f.puts ">#{rec.id}___reverse\n#{rec.seq.reverse}"
-#       end
-#     end
-#
-#     host_fnames.each do |fname|
-#       ParseFasta::SeqFile.open(fname).each_record do |rec|
-#         host_f.puts rec
-#         host_f.puts ">#{rec.id}___reverse\n#{rec.seq.reverse}"
-#       end
-#     end
-#
-#     vir_f.fsync
-#     host_f.fsync
-#
-#     cmd = "mummer -maxmatch -l 15 #{host_f.path} #{vir_f.path} > /Users/moorer/Desktop/mummer.OUT"
-#     Process.run_and_time_it! "MUMMER", cmd
-#   end
-# end
-#
-# header   = nil
-# hits     = []
-# hit_info = {}
-# virus = nil
-#
-# File.open("/Users/moorer/Desktop/mummer.OUT", "rt").each_line.with_index do |line, idx|
-#   if line.start_with? '>'
-#     virus = line.chomp.sub(/^>/, "").sub(/___reverse$/, "").strip
-#
-#     unless hit_info.has_key? virus
-#       hit_info[virus] = {}
-#     end
-#   else
-#     host, _, _, len = line.chomp.strip.split(" ")
-#     host = host.sub(/___reverse$/, "").strip
-#
-#     unless hit_info[virus].has_key? host
-#       hit_info[virus][host] = -1
-#     end
-#
-#     hit_info[virus][host] = len.to_i if len.to_i > hit_info[virus][host]
-#   end
-# end
-#
-# puts
-#
-# hh = hit_info.map do |virus, info|
-#   [virus, info.to_a.sort_by {|gen, len| len}.reverse]
-# end
-#
-# pp hh
-
-# hh = hit_info.map do |virus, info|
-#   [virus, info.to_a.sort_by { |host, hit_len| hit_len }.reverse
-#
-# end
-# p hit_info
-
-scores_files = {}
-programs.each do |program|
-  raw_fname    = File.join outdir, "scores_raw.#{program}.txt"
-  scaled_fname = File.join outdir, "scores_scaled.#{program}.txt"
-
-  scores_files[program] = {
-    raw:    File.open(raw_fname, "w"),
-    scaled: File.open(scaled_fname, "w")
-  }
-end
-
-scores_files.each do |program, files|
-  files.each do |name, file|
-    file.puts %w[virus host score].join "\t"
-  end
-end
-
 name_map_virus, all_ids_virus = BigSimon::Utils.set_up_tmp_dirs virus_fnames, tmpdir_virus, "virus"
 name_map_host, all_ids_host   = BigSimon::Utils.set_up_tmp_dirs host_fnames, tmpdir_host, "host"
 
 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
 
+puts "mummer"
+pp host_info_mummer
+puts
+
+# TODO separate the parser from the runner for homology
+host_info_homology = BigSimon::Runners.homology tmpdir_virus, tmpdir_host, tmpdir, threads
+
+puts "homology"
+pp host_info_homology
+puts
+
+
+
 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], programs
+puts "wish"
+pp host_info_wish
+puts
+
+puts "vhm"
+pp host_info_vhm
+puts
+
+
+host_info_simple_names = BigSimon::Pipeline.collate_host_results [host_info_wish, host_info_vhm, host_info_mummer, host_info_homology], programs
 host_info              = BigSimon::Pipeline.map_taxa host_info_simple_names, name_map_virus, name_map_host
 
 puts
+pp host_info_simple_names
+puts
+
+puts
 pp host_info
 puts
 
 # Just a basic all info file
 # File.open all_predictions_fname, "w" do |f|
 #   f.puts %w[virus host program score scaled.score].join "\t"
 
+scores_files = {}
+programs.each do |program|
+  raw_fname    = File.join outdir, "scores_raw.#{program}.txt"
+  scaled_fname = File.join outdir, "scores_scaled.#{program}.txt"
+
+  scores_files[program] = {
+    raw:    File.open(raw_fname, "w"),
+    scaled: File.open(scaled_fname, "w")
+  }
+end
+
+scores_files.each do |program, files|
+  files.each do |name, file|
+    puts "LALA 3: #{[program, name, file]} #{%w[virus host score].join}"
+
+    file.puts %w[virus host score].join "\t"
+  end
+end
+
 host_info.each do |virus, h1|
   h1.each do |host, h2|
     lines = {}
 
     h2[:scores].each do |program, score|
       lines[[virus, host, program]] = [score]
 
+      puts "LALA: #{[virus, host, score]}"
+
       scores_files[program][:raw].puts [virus, host, score].join "\t"
     end
 
     # Add in the scaled score too.
     h2[:scaled_scores].each do |program, score|
       lines[[virus, host, program]] << score
+
+      puts "LALA 2: #{[virus, host, score]}"
+
 
       scores_files[program][:scaled].puts [virus, host, score].join "\t"
     end
 
     # lines.each do |(virus, host, program), (score, scaled_score)|