miga-base 0.2.4.2 → 0.2.4.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 0568af20da0be58fdf630b68d78851117f5be7fd
-  data.tar.gz: 3941417241eff3bb78965b95a415497ffc09109b
+  metadata.gz: fdd460c0009e55ffc82547e7d31f94789a874442
+  data.tar.gz: c3e560fb2e8871ac586ef8b1fd64c2b1b2076471
 SHA512:
-  metadata.gz: d15f8cb3bec1a804fe09901f52d9d1aedd557136675ea2e9edead1090c0752bde92d36de1414a5d84df799c105afe852eb0d7e4afe3cfb122e3023353333c654
-  data.tar.gz: 1d0527e5f4942560d7efea9cfa17c5f95979ae2b342aca954d859d021fdf7c31d46055065860384706adb05b37712cd0a60d127d90080ce25ba17d1797287574
+  metadata.gz: c77bb9fc35f046a2ebc1c1ddc457f19c4accafe06958b43d42a3654f91a5c19dcf92cc3624f9b30b950542d736787dc69f2e8e9ba88141d20923823da0812f46
+  data.tar.gz: ecee7c4611a74fd16115260e706ea19315ccd639070482010f0bf5c4f669fe30fe9dea755b912053136366831f715cf346f2a32ed38dbee1ea69442525f9e080

data/lib/miga/common.rb

@@ -3,6 +3,7 @@
 
 require "miga/version"
 require "json"
+require "tempfile"
 
 ##
 # Generic class used to handle system-wide information and methods, and parent
@@ -75,6 +76,35 @@ class MiGA::MiGA
     end
   end
 
+  ##
+  # Cleans a FastA file in place.
+  def self.clean_fasta_file(file)
+    tmp = Tempfile.new("MiGA")
+    begin
+      File.open(file, "r") do |fh|
+        buffer = ""
+        fh.each_line do |ln|
+          ln.chomp!
+          if ln =~ /^>\s*(\S+)(.*)/
+            (id, df) = [$1, $2]
+            tmp.print buffer.wrap_width(80)
+            buffer = ""
+            tmp.puts ">#{id.gsub(/[^A-Za-z0-9_\|\.]/, "_")}#{df}"
+          else
+            buffer += ln.gsub(/[^A-Za-z\.\-]/, "")
+          end
+        end
+        tmp.print buffer.wrap_width(80)
+      end
+      tmp.close
+      FileUtils.cp(tmp.path, file)
+    ensure
+      tmp.close
+      tmp.unlink
+    end
+  end
+
+
   ##
   # Check if the result files exist with +base+ name (String) followed by the
   # +ext+ values (Array of String).
@@ -129,4 +159,8 @@ class String
   # Replace underscores by spaces.
   def unmiga_name ; tr("_", " ") ; end
   
+  ##
+  # Wraps the string with fixed Integer +width+.
+  def wrap_width(width) ; gsub(/([^\n\r]{1,#{width}})/,"\\1\n") ; end
+
 end

data/lib/miga/dataset_result.rb

@@ -10,7 +10,7 @@ module MiGA::DatasetResult
     def add_result_raw_reads(base)
       return nil unless result_files_exist?(base, ".1.fastq")
       r = MiGA::Result.new(base + ".json")
-      add_files_to_ds_result(r, name,
+      r = add_files_to_ds_result(r, name,
         ( result_files_exist?(base, ".2.fastq") ?
           {:pair1=>".1.fastq", :pair2=>".2.fastq"} :
           {:single=>".1.fastq"} ))
@@ -62,6 +62,10 @@ module MiGA::DatasetResult
       r = MiGA::Result.new(base + ".json")
       r = add_files_to_ds_result(r, name, {:largecontigs=>".LargeContigs.fna",
         :allcontigs=>".AllContigs.fna", :assembly_data=>""})
+      unless r.clean?
+        MiGA::MiGA.clean_fasta_file(r.file_path :largecontigs)
+        r.clean!
+      end
       add_result(:trimmed_fasta) #-> Post interposing
       r
     end
@@ -71,8 +75,14 @@ module MiGA::DatasetResult
     def add_result_cds(base)
       return nil unless result_files_exist?(base, %w[.faa .fna])
       r = MiGA::Result.new(base + ".json")
-      add_files_to_ds_result(r, name, {:proteins=>".faa", :genes=>".fna",
+      r = add_files_to_ds_result(r, name, {:proteins=>".faa", :genes=>".fna",
         :gff2=>".gff2", :gff3=>".gff3", :tab=>".tab"})
+      unless r.clean?
+        MiGA::MiGA.clean_fasta_file(r.file_path :proteins)
+        MiGA::MiGA.clean_fasta_file(r.file_path :genes)
+        r.clean!
+      end
+      r
     end
 
     ##
@@ -80,7 +90,7 @@ module MiGA::DatasetResult
     def add_result_essential_genes(base)
       return nil unless result_files_exist?(base, %w[.ess.faa .ess .ess/log])
       r = MiGA::Result.new(base + ".json")
-      add_files_to_ds_result(r, name, {:ess_genes=>".ess.faa",
+      r = add_files_to_ds_result(r, name, {:ess_genes=>".ess.faa",
         :collection=>".ess", :report=>".ess/log"})
     end
 
@@ -90,8 +100,13 @@ module MiGA::DatasetResult
       return MiGA::Result.new(base + ".json") if result(:assembly).nil?
       return nil unless result_files_exist?(base, ".ssu.fa")
       r = MiGA::Result.new(base + ".json")
-      add_files_to_ds_result(r, name, {:longest_ssu_gene=>".ssu.fa",
+      r = add_files_to_ds_result(r, name, {:longest_ssu_gene=>".ssu.fa",
         :gff=>".ssu.gff", :all_ssu_genes=>".ssu.all.fa"})
+      unless r.clean?
+        MiGA::MiGA.clean_fasta_file(r.file_path :longest_ssu_gene)
+        r.clean!
+      end
+      r
     end
 
     ##
@@ -103,7 +118,7 @@ module MiGA::DatasetResult
         add_files_to_ds_result(r, name, {:mytaxa=>".mytaxa", :blast=>".blast",
           :mytaxain=>".mytaxain"})
       else
-        MiGA::Result.new base + ".json"
+        MiGA::Result.new(base + ".json")
       end
     end
 
@@ -140,7 +155,7 @@ module MiGA::DatasetResult
     ##
     # Add result type +:stats+ at +base+.
     def add_result_stats(base)
-      MiGA::Result.new(base + ".json")
+      MiGA::Result.new "#{base}.json"
     end
     
     ##
@@ -168,7 +183,7 @@ module MiGA::DatasetResult
         result_files_exist?(base, %w[.aai-medoids.tsv .aai.db]) or
         result_files_exist?(base, %w[.ani-medoids.tsv .ani.db])
       r = MiGA::Result.new(base + ".json")
-      add_files_to_ds_result(r, name, {
+      r = add_files_to_ds_result(r, name, {
         :aai_medoids=>".aai-medoids.tsv",
         :haai_db=>".haai.db", :aai_db=>".aai.db",
         :ani_medoids=>".ani-medoids.tsv", :ani_db=>".ani.db"})

data/lib/miga/project.rb

@@ -272,7 +272,7 @@ class MiGA::Project < MiGA::MiGA
     return MiGA::Result.load(base + ".json") unless save
     return nil unless result_files_exist?(base, ".done")
     r = send("add_result_#{name}", base)
-    r.save
+    r.save unless r.nil?
     r
   end
   

data/lib/miga/project_result.rb

@@ -20,8 +20,7 @@ module MiGA::ProjectResult
     end
 
     def add_result_clade_finding(base)
-      return nil unless result_files_exist?(base,
-        %w[.proposed-clades])
+      return nil unless result_files_exist?(base, %w[.proposed-clades])
       return nil unless is_clade? or result_files_exist?(base,
         %w[.pdf .classif .medoids .class.tsv .class.nwk])
       r = add_result_iter_clades(base)

data/lib/miga/result.rb

@@ -41,6 +41,14 @@ class MiGA::Result < MiGA::MiGA
     @path = path
     MiGA::Result.exist?(path) ? self.load : create
   end
+  
+  ##
+  # Is the result clean? Returns Boolean.
+  def clean? ; !! self[:clean] ; end
+
+  ##
+  # Register the result as cleaned.
+  def clean! ; self[:clean] = true ; end
 
   ##
   # Directory containing the result.

data/lib/miga/version.rb

@@ -10,7 +10,7 @@ module MiGA
   # - Float representing the major.minor version.
   # - Integer representing gem releases of the current version.
   # - Integer representing minor changes that require new version number.
-  VERSION = [0.2, 4, 2]
+  VERSION = [0.2, 4, 3]
   
   ##
   # Nickname for the current major.minor version.
@@ -18,7 +18,7 @@ module MiGA
   
   ##
   # Date of the current gem release.
-  VERSION_DATE = Date.new(2017, 3, 24)
+  VERSION_DATE = Date.new(2017, 3, 28)
   
   ##
   # Reference of MiGA.

data/test/project_test.rb

@@ -75,4 +75,60 @@ class ProjectTest < Test::Unit::TestCase
       File.expand_path("metadata/#{d1.name}.json", p2.path)))
   end
 
+  def test_add_result
+    p1 = $p1
+    assert_nil(p1.add_result(:doom))
+    %w[.Rdata .log .txt .done].each do |x|
+      assert_nil(p1.add_result(:haai_distances))
+      FileUtils.touch(
+        File.expand_path("data/09.distances/01.haai/miga-project#{x}",p1.path))
+    end
+    assert_equal(MiGA::Result, p1.add_result(:haai_distances).class)
+  end
+
+  def test_preprocessing
+    p1 = $p1
+    assert(p1.done_preprocessing?)
+    d1 = p1.add_dataset("BAH")
+    assert(! p1.done_preprocessing?)
+    FileUtils.touch(File.expand_path("data/90.stats/#{d1.name}.done", p1.path))
+    assert(p1.done_preprocessing?)
+    # Distances
+    [:haai_distances, :aai_distances, :ani_distances].each do |r|
+      assert_equal(Symbol, p1.next_distances.class)
+      d = MiGA::Project.RESULT_DIRS[r]
+      %w[.done .Rdata .log .txt].each do |x|
+        assert_nil(p1.add_result(r),
+          "Premature registration of result #{r} at extension #{x}.")
+        FileUtils.touch(File.expand_path("data/#{d}/miga-project#{x}", p1.path))
+      end
+      assert_equal(MiGA::Result, p1.add_result(r).class,
+        "Imposible to add #{r} result.")
+    end
+    assert_equal(:clade_finding, p1.next_distances)
+    
+    # Clades
+    assert_nil(p1.next_inclade)
+    p1.metadata[:type] = :clade
+    res = [
+      [:clade_finding,
+        %w[.pdf .classif .medoids .class.tsv .class.nwk .proposed-clades]],
+      [:subclades, %w[.pdf .classif .medoids .class.tsv .class.nwk]],
+      [:ogs, %w[.ogs .stats]]
+    ]
+    res.each do |rr|
+      (r, xs) = rr
+      d = MiGA::Project.RESULT_DIRS[r]
+      assert_equal(Symbol, p1.next_inclade.class)
+      ([".done"] + xs).each do |x|
+        assert_nil(p1.add_result(r),
+          "Premature registration of result #{r} at extension #{x}.")
+        FileUtils.touch(File.expand_path("data/#{d}/miga-project#{x}", p1.path))
+      end
+      assert_equal(MiGA::Result, p1.add_result(r).class,
+        "Impossible to add #{r} result.")
+    end
+    assert_nil(p1.next_inclade)
+  end
+
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: miga-base
 version: !ruby/object:Gem::Version
-  version: 0.2.4.2
+  version: 0.2.4.3
 platform: ruby
 authors:
 - Luis M. Rodriguez-R
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2017-03-24 00:00:00.000000000 Z
+date: 2017-03-28 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rest-client