big_simon 0.2.0 → 0.2.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 95ca94aff107e2545dd26ab671c2ed125cb89164
-  data.tar.gz: 8cb9098b4462d6702ecd692e7261abb9c1c7c26a
+  metadata.gz: 624fd8e70d43c1a556b9c846e599dd928cd0c081
+  data.tar.gz: 456b7aec02cd33c6d0145d144fb83504daa0ef07
 SHA512:
-  metadata.gz: be890e07e1ca248da8e5b80637b25d00c20185d4710229cc9c72d564abc930d4eb734ad53f42842a479827e64205ee1f53f5691b6527d03b9cc41b9b8656ae85
-  data.tar.gz: ac30f597de91303e3b8040ecfde48f98d56c343aadfa3a1085d77fbb55692fb055569734470812144d716e73ee8c94265c732a07462bd8490c2b937572961062
+  metadata.gz: 431d7b1f0f44af912b42adf458df8a15ff9b60139b5c6c27318e442b00a55f94b289b59e4518f9005b52d5b26de712ea2bd142c8a152558a98c47e4ede7a2a82
+  data.tar.gz: 6282c6ce73c86f4276934760c4b9decf2d758d962b5c43b7d84048e420fda41eee97d9c0fe1a681479f93daee16a1545f79fee56aba41e0fd0d06eec0debf9ab

data/.gitignore

@@ -38,4 +38,6 @@ spec/test_files/homology_files/output/*homology*
 
 pec/test_files/mummer_files/output/mummer_out.txt
 
-tmp.prodigal.stdin.5129
+tmp.prodigal.stdin.5129
+
+spec/test_files/mummer_files/output/

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    big_simon (0.2.0)
+    big_simon (0.2.1)
       parallel
       parse_fasta (~> 2.5, >= 2.5.2)
       rya (~> 0.4.0)

data/exe/big_simon

@@ -73,8 +73,10 @@ tmpdir_host  = File.join tmpdir, "host"
 mean_scaled_scores_fname = File.join outdir, "scores_scaled.mean.txt"
 
 
-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"
+name_map_virus, all_ids_virus, vir_seq_lengths = BigSimon::Utils.set_up_tmp_dirs virus_fnames, tmpdir_virus, "virus"
+name_map_host, all_ids_host, host_seq_lengths   = BigSimon::Utils.set_up_tmp_dirs host_fnames, tmpdir_host, "host"
+
+all_seq_lengths = vir_seq_lengths.merge host_seq_lengths
 
 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

data/lib/big_simon/runners.rb

@@ -6,7 +6,7 @@ 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
+    def self.mummer exe, vir_dir, host_dir, outdir, threads, all_seq_lengths
       klass = Class.new.extend Rya::CoreExtensions::Math
       FileUtils.mkdir_p outdir
 
@@ -80,10 +80,17 @@ module BigSimon
             ary = line.strip.split " "
 
             host  = ary[0].sub(/___reverse$/, "").strip
-            score = ary[3].to_i
 
             Rya::AbortIf.assert hit_table[virus].has_key?(host)
 
+            Rya::AbortIf.assert all_seq_lengths[virus]
+            Rya::AbortIf.assert all_seq_lengths[host]
+
+            combined_seq_length = all_seq_lengths[virus] + all_seq_lengths[host]
+
+            score = ary[3].to_i / combined_seq_length * 1000
+
+
             # unless hit_table[virus].has_key? host
             #   hit_table[virus][host] = -1
             # end
@@ -213,7 +220,7 @@ module BigSimon
     # @note Assumes that the files end with *.fa
     # @note Assumes that the file names match the IDs.  This SHOULD be taken care of by the big_simon program.
     # @todo assert that fname thing matches sequence ID name.
-    def self.homology vir_dir, host_dir, outdir, threads
+    def self.homology vir_dir, host_dir, outdir, threads, all_seq_lengths
       FileUtils.mkdir_p outdir
 
       host_orfs          = File.join outdir, "host_orfs.homology"
@@ -272,6 +279,13 @@ module BigSimon
 
           Rya::AbortIf.assert blast_table.has_key?(vir_id), "blast_table: got #{vir_id} should have been #{vir_simple_fname}"
 
+          Rya::AbortIf.assert all_seq_lengths[vir_id]
+          Rya::AbortIf.assert all_seq_lengths[host_id]
+
+          combined_seq_length = all_seq_lengths[vir_id] + all_seq_lengths[host_id]
+          score = score / combined_seq_length.to_f * 1000
+
+
           blast_table[vir_id][host_id] += score
         end
 
@@ -309,7 +323,10 @@ module BigSimon
           collated_blast_table[vir_id] = []
 
           host_simple_names.each do |host_id|
-            scaled_score = klass.scale host_scores[host_id], 0, max_score, 1, 0
+
+
+            combined_seq_length = all_seq_lengths[vir_id] + all_seq_lengths[host_id]
+            scaled_score = klass.scale host_scores[host_id].to_f, 0, max_score, 1, 0
 
             host_table = { host: host_id, score: host_scores[host_id], scaled_score: scaled_score }
             collated_blast_table[vir_id] << host_table

data/lib/big_simon/utils.rb

@@ -74,6 +74,8 @@ draw.heatmap("#{in_fname}", "#{out_fname}")
     def self.set_up_tmp_dirs fastas, tmpdir, which
       Object::FileUtils.mkdir_p tmpdir
 
+      seq_lengths = {}
+
       name_map = {}
       all_ids  = Set.new
 
@@ -92,6 +94,8 @@ draw.heatmap("#{in_fname}", "#{out_fname}")
 
             outfname = File.join tmpdir, "#{new_id}.fa"
 
+            seq_lengths[new_id] = rec.seq.length
+
             File.open(outfname, "w") do |f|
               f.puts ">#{new_id}\n#{rec.seq}" # TODO HERE
             end
@@ -99,7 +103,7 @@ draw.heatmap("#{in_fname}", "#{out_fname}")
         end
       end
 
-      [name_map, all_ids]
+      [name_map, all_ids, seq_lengths]
     end
 
     def self.strip_suffix fname

data/lib/big_simon/version.rb

@@ -1,5 +1,5 @@
 module BigSimon
-  VERSION = "0.2.0"
+  VERSION = "0.2.1"
 
   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: 0.2.0
+  version: 0.2.1
 platform: ruby
 authors:
 - Ryan Moore