miga-base 0.2.0.9 → 0.2.1.0

data/lib/miga/dataset.rb

@@ -3,11 +3,14 @@
 
 require "miga/metadata"
 require "miga/result"
+require "miga/dataset_result"
 
 ##
 # Dataset representation in MiGA.
 class MiGA::Dataset < MiGA::MiGA
   
+  include MiGA::DatasetResult
+  
   # Class-level
 
   ##
@@ -52,7 +55,7 @@ class MiGA::Dataset < MiGA::MiGA
   
   ##
   # Tasks to be excluded from query datasets.
-  @@EXCLUDE_NOREF_TASKS = [:essential_genes, :mytaxa_scan]
+  @@EXCLUDE_NOREF_TASKS = [:mytaxa_scan]
   
   ##
   # Tasks to be executed only in datasets that are not multi-organism. These
@@ -137,14 +140,14 @@ class MiGA::Dataset < MiGA::MiGA
   # Is this dataset known to be multi-organism?
   def is_multi?
     return false if self.metadata[:type].nil?
-    return @@KNOWN_TYPES[self.metadata[:type]][:multi]
+    @@KNOWN_TYPES[self.metadata[:type]][:multi]
   end
   
   ##
   # Is this dataset known to be single-organism?
   def is_nonmulti?
     return false if self.metadata[:type].nil?
-    return !@@KNOWN_TYPES[self.metadata[:type]][:multi]
+    !@@KNOWN_TYPES[self.metadata[:type]][:multi]
   end
   
   ##
@@ -176,7 +179,7 @@ class MiGA::Dataset < MiGA::MiGA
     return nil if @@RESULT_DIRS[result_type].nil?
     base = project.path + "/data/" + @@RESULT_DIRS[result_type] +
       "/" + name
-    return MiGA::Result.load(base + ".json") unless save
+    return MiGA::Result.load("#{base}.json") unless save
     return nil unless result_files_exist?(base, ".done")
     r = self.send("add_result_#{result_type}", base)
     r.save unless r.nil?
@@ -241,123 +244,4 @@ class MiGA::Dataset < MiGA::MiGA
     adv
   end
 
-  private
-    
-    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,
-        ( result_files_exist?(base, ".2.fastq") ?
-          {:pair1=>".1.fastq", :pair2=>".2.fastq"} :
-          {:single=>".1.fastq"} ))
-    end
-
-    def add_result_trimmed_reads(base)
-      return nil unless result_files_exist?(base, ".1.clipped.fastq")
-      r = MiGA::Result.new base + ".json"
-      r = add_files_to_ds_result(r, name,
-        {:pair1=>".1.clipped.fastq", :pair2=>".2.clipped.fastq"}) if
-        result_files_exist?(base, ".2.clipped.fastq")
-      r.add_file(:single, name + ".1.clipped.single.fastq")
-      add_result(:raw_reads) #-> Post gunzip
-      r
-    end
-
-    def add_result_read_quality(base)
-      return nil unless result_files_exist?(base, %w[.solexaqa .fastqc])
-      r = MiGA::Result.new(base + ".json")
-      r = add_files_to_ds_result(r, name,
-        {:solexaqa=>".solexaqa", :fastqc=>".fastqc"})
-      add_result(:trimmed_reads) #-> Post cleaning
-      r
-    end
-
-    def add_result_trimmed_fasta(base)
-      return nil unless
-        result_files_exist?(base, ".CoupledReads.fa") or
-        result_files_exist?(base, ".SingleReads.fa")
-      r = MiGA::Result.new base + ".json"
-      r = add_files_to_ds_result(r, name, {:coupled=>".CoupledReads.fa",
-        :pair1=>".1.fa", :pair2=>".2.fa"}) if
-        result_files_exist?(base, ".CoupledReads.fa")
-      r.add_file(:single, name + ".SingleReads.fa")
-      add_result(:raw_reads) #-> Post gzip
-      r
-    end
-
-    def add_result_assembly(base)
-      return nil unless result_files_exist?(base, ".LargeContigs.fna")
-      r = MiGA::Result.new(base + ".json")
-      add_files_to_ds_result(r, name, {:largecontigs=>".LargeContigs.fna",
-        :allcontigs=>".AllContigs.fna"})
-    end
-
-    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",
-        :gff2=>".gff2", :gff3=>".gff3", :tab=>".tab"})
-    end
-
-    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",
-        :collection=>".ess", :report=>".ess/log"})
-    end
-
-    def add_result_ssu(base)
-      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",
-        :gff=>".ssu.gff", :all_ssu_genes=>".ssu.all.fa"})
-    end
-
-    def add_result_mytaxa(base)
-      if is_multi?
-        return nil unless result_files_exist?(base, ".mytaxa")
-        r = MiGA::Result.new(base + ".json")
-        add_files_to_ds_result(r, name, {:mytaxa=>".mytaxa", :blast=>".blast",
-          :mytaxain=>".mytaxain"})
-      else
-        MiGA::Result.new base + ".json"
-      end
-    end
-
-    def add_result_mytaxa_scan(base)
-      if is_nonmulti?
-        return nil unless
-          result_files_exist?(base, %w[.pdf .wintax .mytaxa .reg])
-        r = MiGA::Result.new(base + ".json")
-        add_files_to_ds_result(r, name, {:mytaxa=>".mytaxa", :wintax=>".wintax",
-          :blast=>".blast", :mytaxain=>".mytaxain", :report=>".pdf",
-          :regions=>".reg", :gene_ids=>".wintax.genes",
-          :region_ids=>".wintax.regions"})
-      else
-        MiGA::Result.new base + ".json"
-      end
-    end
-
-    def add_result_distances(base)
-      if is_nonmulti?
-        pref = File.dirname(base)
-        return nil unless
-          File.exist?("#{pref}/#{is_ref? ? "01.haai" : "02.aai"}/#{name}.db")
-        r = MiGA::Result.new(base + ".json")
-        r.add_files({:haai_db=>"01.haai/#{name}.db",
-          :aai_db=>"02.aai/#{name}.db", :ani_db=>"03.ani/#{name}.db"})
-      else
-        r = MiGA::Result.new "#{base}.json"
-      end
-      r
-    end
-
-    def add_files_to_ds_result(r, name, rel_files)
-      files = {}
-      rel_files.each{ |k,v| files[k] = name + v }
-      r.add_files(files)
-      r
-    end
-
 end # class MiGA::Dataset

data/lib/miga/dataset_result.rb

@@ -0,0 +1,177 @@
+
+##
+# Helper module including specific functions to add dataset results.
+module MiGA::DatasetResult
+  
+  private
+
+    ##
+    # Add result type +:raw_reads+ at +base+.
+    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,
+        ( result_files_exist?(base, ".2.fastq") ?
+          {:pair1=>".1.fastq", :pair2=>".2.fastq"} :
+          {:single=>".1.fastq"} ))
+    end
+
+    ##
+    # Add result type +:trimmed_reads+ at +base+.
+    def add_result_trimmed_reads(base)
+      return nil unless result_files_exist?(base, ".1.clipped.fastq")
+      r = MiGA::Result.new base + ".json"
+      r = add_files_to_ds_result(r, name,
+        {:pair1=>".1.clipped.fastq", :pair2=>".2.clipped.fastq"}) if
+        result_files_exist?(base, ".2.clipped.fastq")
+      r.add_file(:single, name + ".1.clipped.single.fastq")
+      add_result(:raw_reads) #-> Post gunzip
+      r
+    end
+
+    ##
+    # Add result type +:read_quality+ at +base+.
+    def add_result_read_quality(base)
+      return nil unless result_files_exist?(base, %w[.solexaqa .fastqc])
+      r = MiGA::Result.new(base + ".json")
+      r = add_files_to_ds_result(r, name,
+        {:solexaqa=>".solexaqa", :fastqc=>".fastqc"})
+      add_result(:trimmed_reads) #-> Post cleaning
+      r
+    end
+
+    ##
+    # Add result type +:trimmed_fasta+ at +base+.
+    def add_result_trimmed_fasta(base)
+      return nil unless
+        result_files_exist?(base, ".CoupledReads.fa") or
+        result_files_exist?(base, ".SingleReads.fa") or
+        result_files_exist?(base, %w[.1.fasta .2.fasta])
+      r = MiGA::Result.new base + ".json"
+      r = add_files_to_ds_result(r, name, {:coupled=>".CoupledReads.fa",
+        :single=>".SingleReads.fa", :pair1=>".1.fasta", :pair2=>".2.fasta"})
+      add_result(:raw_reads) #-> Post gzip
+      r
+    end
+
+    ##
+    # Add result type +:assembly+ at +base+.
+    def add_result_assembly(base)
+      return nil unless result_files_exist?(base, ".LargeContigs.fna")
+      r = MiGA::Result.new(base + ".json")
+      add_files_to_ds_result(r, name, {:largecontigs=>".LargeContigs.fna",
+        :allcontigs=>".AllContigs.fna"})
+    end
+
+    ##
+    # Add result type +:cds+ at +base+.
+    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",
+        :gff2=>".gff2", :gff3=>".gff3", :tab=>".tab"})
+    end
+
+    ##
+    # Add result type +:essential_genes+ at +base+.
+    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",
+        :collection=>".ess", :report=>".ess/log"})
+    end
+
+    ##
+    # Add result type +:ssu+ at +base+.
+    def add_result_ssu(base)
+      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",
+        :gff=>".ssu.gff", :all_ssu_genes=>".ssu.all.fa"})
+    end
+
+    ##
+    # Add result type +:mytaxa+ at +base+.
+    def add_result_mytaxa(base)
+      if is_multi?
+        return nil unless result_files_exist?(base, ".mytaxa")
+        r = MiGA::Result.new(base + ".json")
+        add_files_to_ds_result(r, name, {:mytaxa=>".mytaxa", :blast=>".blast",
+          :mytaxain=>".mytaxain"})
+      else
+        MiGA::Result.new base + ".json"
+      end
+    end
+
+    ##
+    # Add result type +:mytaxa_scan+ at +base+.
+    def add_result_mytaxa_scan(base)
+      if is_nonmulti?
+        return nil unless
+          result_files_exist?(base, %w[.pdf .wintax .mytaxa .reg])
+        r = MiGA::Result.new(base + ".json")
+        add_files_to_ds_result(r, name, {:mytaxa=>".mytaxa", :wintax=>".wintax",
+          :blast=>".blast", :mytaxain=>".mytaxain", :report=>".pdf",
+          :regions=>".reg", :gene_ids=>".wintax.genes",
+          :region_ids=>".wintax.regions"})
+      else
+        MiGA::Result.new base + ".json"
+      end
+    end
+
+    ##
+    # Add result type +:distances+ at +base+.
+    def add_result_distances(base)
+      if is_nonmulti?
+        if is_ref?
+          add_result_distances_ref(base)
+        else
+          add_result_distances_nonref(base)
+        end
+      else
+        add_result_distances_multi(base)
+      end
+    end
+    
+    ##
+    # Add result type +:distances+ for _multi_ datasets at +base+.
+    def add_result_distances_multi(base)
+      MiGA::Result.new "#{base}.json"
+    end
+
+    ##
+    # Add result type +:distances+ for _nonmulti_ reference datasets at +base+.
+    def add_result_distances_ref(base)
+      pref = File.dirname(base)
+      return nil unless
+        File.exist?("#{pref}/01.haai/#{name}.db")
+      r = MiGA::Result.new(base + ".json")
+      r.add_files({:haai_db=>"01.haai/#{name}.db",
+        :aai_db=>"02.aai/#{name}.db", :ani_db=>"03.ani/#{name}.db"})
+      r
+    end
+
+    ##
+    # Add result type +:distances+ for _nonmulti_ query datasets at +base+.
+    def add_result_distances_nonref(base)
+      return nil unless
+        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, {
+        :aai_medoids=>".aai-medoids.tsv",
+        :haai_db=>".haai.db", :aai_db=>".aai.db",
+        :ani_medoids=>".ani-medoids.tsv", :ani_db=>".ani.db"})
+    end
+
+    ##
+    # Add files in +rel_files+ Hash to the result +r+ with dataset name +name+.
+    def add_files_to_ds_result(r, name, rel_files)
+      files = {}
+      rel_files.each{ |k,v| files[k] = name + v }
+      r.add_files(files)
+      r
+    end
+
+end

data/lib/miga/project.rb

@@ -146,12 +146,26 @@ class MiGA::Project < MiGA::MiGA
   ##
   # Name of the project.
   def name ; metadata[:name] ; end
+
+  ##
+  # Type of project.
+  def type ; metadata[:type] ; end
+
+  ##
+  # Is this a clade project?
+  def is_clade? ; type==:clade ; end
   
   ##
   # Returns Array of MiGA::Dataset.
   def datasets
     metadata[:datasets].map{ |name| dataset(name) }
   end
+
+  ##
+  # Returns Array of String (without evaluating dataset objects).
+  def dataset_names
+    metadata[:datasets]
+  end
   
   ##
   # Returns MiGA::Dataset.
@@ -164,17 +178,24 @@ class MiGA::Project < MiGA::MiGA
   end
   
   ##
-  # Iterate through datasets, with a single variable MiGA::Dataset passed to
-  # +blk+.
+  # Iterate through datasets, with one or two variables passed to +blk+.
+  # If one, the dataset MiGA::Dataset object is passed. If two, the name and
+  # the dataset object are passed.
   def each_dataset(&blk)
-    metadata[:datasets].each{ |name| blk.call(dataset(name)) }
+    metadata[:datasets].each do |name|
+      if blk.arity == 1
+        blk.call(dataset(name))
+      else
+        blk.call(name, dataset(name))
+      end
+    end
   end
   
   ##
   # Add dataset identified by +name+ and return MiGA::Dataset.
   def add_dataset(name)
     unless metadata[:datasets].include? name
-      d = MiGA::Dataset.new(self, name)
+      MiGA::Dataset.new(self, name)
       @metadata[:datasets] << name
       save
     end
@@ -324,7 +345,9 @@ class MiGA::Project < MiGA::MiGA
 
     def add_result_clade_finding(base)
       return nil unless result_files_exist?(base,
-        %w[.proposed-clades .pdf .1.classif .1.medoids .class.tsv .class.nwk])
+        %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)
       r.add_file(:aai_tree,	"miga-project.aai.nwk")
       r.add_file(:proposal,	"miga-project.proposed-clades")
@@ -335,9 +358,9 @@ class MiGA::Project < MiGA::MiGA
 
     def add_result_subclades(base)
       return nil unless result_files_exist?(base,
-        %w[.pdf .1.classif .1.medoids .class.tsv .class.nwk])
+        %w[.pdf .classif .medoids .class.tsv .class.nwk])
       r = add_result_iter_clades(base)
-      r.add_file(:ani_tree,	"miga-project.ani.nwk")
+      r.add_file(:ani_tree, "miga-project.ani.nwk")
       r
     end
 
@@ -346,11 +369,8 @@ class MiGA::Project < MiGA::MiGA
       r.add_file(:report,	"miga-project.pdf")
       r.add_file(:class_table,	"miga-project.class.tsv")
       r.add_file(:class_tree,	"miga-project.class.nwk")
-      (1..6).each do |i|
-        %w{classif medoids}.each do |m|
-          r.add_file("#{m}_#{i}".to_sym, "miga-project.#{i}.#{m}")
-        end
-      end
+      r.add_file(:classif,	"miga-project.classif")
+      r.add_file(:medoids,	"miga-project.medoids")
       r
     end
 

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, 0, 9]
+  VERSION = [0.2, 1, 0]
   
   ##
   # Nickname for the current major.minor version.
@@ -18,7 +18,7 @@ module MiGA
   
   ##
   # Date of the current gem release.
-  VERSION_DATE = Date.new(2016, 04, 13)
+  VERSION_DATE = Date.new(2016, 04, 20)
   
   ##
   # Reference of MiGA.