miga-base 0.6.2.0 → 0.6.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 2c317237c4cd2f049b56af6021272208c206f1810052dde4b1b5b41394f37139
-  data.tar.gz: '066558242d9bcc1a96f166349815398296cdb976190853a18ca52bdafe84c263'
+  metadata.gz: 8b225951f374bcd267560e5bd8234fb88bcd6b0c11b0561fb4b3b479af39c4b3
+  data.tar.gz: 9b32d40ea94ceb526fe0ba732c77fce978b0cba5decffd4e1c0d701594670dbd
 SHA512:
-  metadata.gz: 50b395740f68543e7aa79bf0e3631cb86248fffbcd9aac942e2d30681579a42e528d552638701f153605f68ea9da389762b6d068455e9b67cbfd9b66cde879c0
-  data.tar.gz: d20fd84e71bd4d7a8ebf5ff1060e12096d78477b662098e013b88ea9a19e54b4665a71bbfa0e49fd57df47a61b3a0780ed469fcb004b1384dd4e4f2304320666
+  metadata.gz: b7fe9b2cbb09b6612b762c7c9202b4b27ece7a0b6f4dd23eecee9bddc835c130f15a63772011106a1f0f1425e5445fa1541a8ebc81661ef341dcacec3ae22193
+  data.tar.gz: 501fa797aa6726ac5cdc6c043a3073d03a7ba3ed81d63ebac9ef989a76aa1947806c9052ab50c089567001ad1b1b92ad04cc05ea4ca33691de36c4dcb2e34b52

data/lib/miga/cli/action/new.rb

@@ -11,20 +11,25 @@ class MiGA::Cli::Action::New < MiGA::Cli::Action
       opt.on(
         '-n', '--name STRING',
         'Name of the project'
-        ){ |v| cli[:name] = v }
+      ) { |v| cli[:name] = v }
       opt.on(
         '-d', '--description STRING',
         'Description of the project'
-        ){ |v| cli[:description] = v }
+      ) { |v| cli[:description] = v }
       opt.on(
         '-c', '--comments STRING',
         'Comments on the project'
-        ){ |v| cli[:comments] = v }
+      ) { |v| cli[:comments] = v }
+      opt.on(
+        '--fast',
+        'Use faster identity engines (Diamond-AAI and FastANI)',
+        'Equivalent to: -m aai_p=diamond,ani_p=fastani'
+      ) { |v| cli[:fast] = v }
       opt.on(
         '-m', '--metadata STRING',
         'Metadata as key-value pairs separated by = and delimited by comma',
         'Values are saved as strings except for booleans (true / false) or nil'
-        ){ |v| cli[:metadata] = v }
+      ) { |v| cli[:metadata] = v }
     end
   end
 
@@ -40,6 +45,10 @@ class MiGA::Cli::Action::New < MiGA::Cli::Action
     raise 'Project already exists, aborting.' if Project.exist? cli[:project]
     p = Project.new(cli[:project], false)
     p = cli.add_metadata(p)
+    if cli[:fast]
+      p.metadata[:aai_p] = 'diamond'
+      p.metadata[:ani_p] = 'fastani'
+    end
     p.save
   end
 end

data/lib/miga/dataset.rb

@@ -9,10 +9,9 @@ require 'sqlite3'
 ##
 # Dataset representation in MiGA.
 class MiGA::Dataset < MiGA::MiGA
-  
   include MiGA::Dataset::Result
   include MiGA::Dataset::Hooks
-  
+
   # Class-level
   class << self
 
@@ -27,7 +26,7 @@ class MiGA::Dataset < MiGA::MiGA
     def INFO_FIELDS
       %w(name created updated type ref user description comments)
     end
-  
+
   end
 
   # Instance-level
@@ -35,11 +34,11 @@ class MiGA::Dataset < MiGA::MiGA
   ##
   # MiGA::Project that contains the dataset.
   attr_reader :project
-  
+
   ##
   # Datasets are uniquely identified by +name+ in a project.
   attr_reader :name
-  
+
   ##
   # Create a MiGA::Dataset object in a +project+ MiGA::Project with a
   # uniquely identifying +name+. +is_ref+ indicates if the dataset is to
@@ -52,21 +51,25 @@ class MiGA::Dataset < MiGA::MiGA
     end
     @project = project
     @name = name
+    @metadata = nil
     metadata[:ref] = is_ref
     @metadata_future = [
       File.expand_path("metadata/#{name}.json", project.path),
       metadata
     ]
     save unless File.exist? @metadata_future[0]
-    pull_hook :on_load
   end
 
   ##
-  # MiGA::Metadata with information about the dataset.
+  # MiGA::Metadata with information about the dataset
   def metadata
-    @metadata ||= MiGA::Metadata.new(*@metadata_future)
+    if @metadata.nil?
+      @metadata = MiGA::Metadata.new(*@metadata_future)
+      pull_hook :on_load
+    end
+    @metadata
   end
-  
+
   ##
   # Save any changes you've made in the dataset.
   def save
@@ -74,11 +77,11 @@ class MiGA::Dataset < MiGA::MiGA
     metadata.save
     pull_hook :on_save
   end
-  
+
   ##
   # Get the type of dataset as Symbol.
   def type ; metadata[:type] ; end
-  
+
   ##
   # Delete the dataset with all it's contents (including results) and returns
   # nil.
@@ -103,7 +106,7 @@ class MiGA::Dataset < MiGA::MiGA
     self.metadata.save
     pull_hook :on_activate
   end
-  
+
   ##
   # Get standard metadata values for the dataset as Array.
   def info
@@ -111,7 +114,7 @@ class MiGA::Dataset < MiGA::MiGA
       (k == 'name') ? self.name : metadata[k.to_sym]
     end
   end
-  
+
   ##
   # Is this dataset a reference?
   def is_ref? ; !!metadata[:ref] ; end
@@ -119,14 +122,14 @@ class MiGA::Dataset < MiGA::MiGA
   ##
   # Is this dataset a query (non-reference)?
   def is_query? ; !metadata[:ref] ; end
-  
+
   ##
   # Is this dataset known to be multi-organism?
   def is_multi?
     return false if metadata[:type].nil? or @@KNOWN_TYPES[type].nil?
     @@KNOWN_TYPES[type][:multi]
   end
-  
+
   ##
   # Is this dataset known to be single-organism?
   def is_nonmulti?
@@ -139,7 +142,7 @@ class MiGA::Dataset < MiGA::MiGA
   def is_active?
     metadata[:inactive].nil? or !metadata[:inactive]
   end
-  
+
   ##
   # Should I ignore +task+ for this dataset?
   def ignore_task?(task)

data/lib/miga/dataset/result.rb

@@ -24,10 +24,10 @@ module MiGA::Dataset::Result
   end
 
   ##
-  # For each result executes the 2-ary +blk+ block: key symbol and MiGA::Result
-  def each_result(&blk)
-    @@RESULT_DIRS.keys.each do |k|
-      blk.call(k, result(k)) unless result(k).nil?
+  # For each result executes the 2-ary block: key symbol and MiGA::Result
+  def each_result
+    @@RESULT_DIRS.each_key do |k|
+      yield(k, result(k)) unless result(k).nil?
     end
   end
 
@@ -50,20 +50,20 @@ module MiGA::Dataset::Result
       r_pre = MiGA::Result.load("#{base}.json")
       return r_pre if (r_pre.nil? && !save) || !r_pre.nil?
     end
-    r = if File.exist?("#{base}.done")
-      self.send("add_result_#{result_type}", base, opts)
-    end
-    unless r.nil?
-      r.save
-      pull_hook(:on_result_ready, result_type)
-    end
+    fun = "add_result_#{result_type}"
+    r = send(fun, base, opts) if File.exist?("#{base}.done")
+    return if r.nil?
+    r.save
+    pull_hook(:on_result_ready, result_type)
     r
   end
 
   ##
   # Gets a result as MiGA::Result for the datasets with +result_type+. This is
   # equivalent to +add_result(result_type, false)+.
-  def get_result(result_type) ; add_result(result_type, false) ; end
+  def get_result(result_type)
+    add_result(result_type, false)
+  end
 
   ##
   # Returns the key symbol of the first registered result (sorted by the
@@ -179,7 +179,7 @@ module MiGA::Dataset::Result
     r = get_result(:distances)
     ref = project.datasets.select(&:is_ref?).select(&:is_active?).map(&:name)
     return if r.nil?
-    [:haai_db, :aai_db, :ani_db].each do |db_type|
+    %i[haai_db aai_db ani_db].each do |db_type|
       db = r.file_path(db_type)
       next if db.nil? || !File.size?(db)
       sqlite_db = SQLite3::Database.new db
@@ -198,9 +198,8 @@ module MiGA::Dataset::Result
   # Add result type +:raw_reads+ at +base+ (no +_opts+ supported)
   def add_result_raw_reads(base, _opts)
     return nil unless result_files_exist?(base, '.1.fastq')
-    r = MiGA::Result.new("#{base}.json")
     add_files_to_ds_result(
-      r, name,
+      MiGA::Result.new("#{base}.json"), name,
       if result_files_exist?(base, '.2.fastq')
         { pair1: '.1.fastq', pair2: '.2.fastq' }
       else
@@ -213,25 +212,30 @@ module MiGA::Dataset::Result
   # Add result type +:trimmed_reads+ at +base+ (no +_opts+ supported)
   def add_result_trimmed_reads(base, _opts)
     return nil unless result_files_exist?(base, '.1.clipped.fastq')
-    r = MiGA::Result.new("#{base}.json")
-    if result_files_exist?(base, '.2.clipped.fastq')
-      r = add_files_to_ds_result(r, name,
-        pair1: '.1.clipped.fastq',
-        pair2: '.2.clipped.fastq',
-        single: '.1.clipped.single.fastq')
-    else
-      r = add_files_to_ds_result(r, name, single: '.1.clipped.fastq')
+    add_files_to_ds_result(
+      MiGA::Result.new("#{base}.json"), name,
+      if result_files_exist?(base, '.2.clipped.fastq')
+        {
+          pair1: '.1.clipped.fastq',
+          pair2: '.2.clipped.fastq',
+          single: '.1.clipped.single.fastq'
+        }
+      else
+        { single: '.1.clipped.fastq' }
+      end
+    ).tap do |r|
+      r.add_file(:trimming_sumary, "#{name}.1.fastq.trimmed.summary.txt")
     end
-    r.add_file(:trimming_sumary, "#{name}.1.fastq.trimmed.summary.txt")
-    r
   end
 
   ##
   # Add result type +:read_quality+ at +base+ (no +_opts+ supported)
   def add_result_read_quality(base, _opts)
     return nil unless result_files_exist?(base, %w[.solexaqa .fastqc])
-    r = MiGA::Result.new("#{base}.json")
-    add_files_to_ds_result(r, name, solexaqa: '.solexaqa', fastqc: '.fastqc')
+    add_files_to_ds_result(
+      MiGA::Result.new("#{base}.json"), name,
+      solexaqa: '.solexaqa', fastqc: '.fastqc'
+    )
   end
 
   ##
@@ -241,13 +245,13 @@ module MiGA::Dataset::Result
       result_files_exist?(base, '.CoupledReads.fa') ||
       result_files_exist?(base, '.SingleReads.fa')  ||
       result_files_exist?(base, %w[.1.fasta .2.fasta])
-    r = MiGA::Result.new("#{base}.json")
     add_files_to_ds_result(
-      r, name,
+      MiGA::Result.new("#{base}.json"), name,
       coupled: '.CoupledReads.fa',
       single: '.SingleReads.fa',
       pair1: '.1.fasta',
-      pair2: '.2.fasta')
+      pair2: '.2.fasta'
+    )
   end
 
   ##
@@ -255,12 +259,12 @@ module MiGA::Dataset::Result
   # +is_clean: Boolean+.
   def add_result_assembly(base, opts)
     return nil unless result_files_exist?(base, '.LargeContigs.fna')
-    r = MiGA::Result.new("#{base}.json")
     r = add_files_to_ds_result(
-      r, name,
+      MiGA::Result.new("#{base}.json"), name,
       largecontigs: '.LargeContigs.fna',
       allcontigs: '.AllContigs.fna',
-      assembly_data: '')
+      assembly_data: ''
+    )
     opts[:is_clean] ||= false
     r.clean! if opts[:is_clean]
     unless r.clean?
@@ -274,14 +278,14 @@ module MiGA::Dataset::Result
   # Add result type +:cds+ at +base+. Hash +opts+ supports +is_clean: Boolean+
   def add_result_cds(base, opts)
     return nil unless result_files_exist?(base, %w[.faa])
-    r = MiGA::Result.new("#{base}.json")
     r = add_files_to_ds_result(
-      r, name,
+      MiGA::Result.new("#{base}.json"), name,
       proteins: '.faa',
       genes: '.fna',
       gff2: '.gff2',
       gff3: '.gff3',
-      tab: '.tab')
+      tab: '.tab'
+    )
     opts[:is_clean] ||= false
     r.clean! if opts[:is_clean]
     unless r.clean?
@@ -296,13 +300,13 @@ module MiGA::Dataset::Result
   # Add result type +:essential_genes+ at +base+ (no +_opts+ supported).
   def add_result_essential_genes(base, _opts)
     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,
+      MiGA::Result.new("#{base}.json"), name,
       ess_genes: '.ess.faa',
       collection: '.ess',
       report: '.ess/log',
-      alignments: '.ess/proteins.aln')
+      alignments: '.ess/proteins.aln'
+    )
   end
 
   ##
@@ -310,12 +314,12 @@ module MiGA::Dataset::Result
   def add_result_ssu(base, opts)
     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")
     r = add_files_to_ds_result(
-      r, name,
+      MiGA::Result.new("#{base}.json"), name,
       longest_ssu_gene: '.ssu.fa',
       gff: '.ssu.gff',
-      all_ssu_genes: '.ssu.all.fa')
+      all_ssu_genes: '.ssu.all.fa'
+    )
     opts[:is_clean] ||= false
     r.clean! if opts[:is_clean]
     unless r.clean?
@@ -332,9 +336,8 @@ module MiGA::Dataset::Result
       return nil unless
         result_files_exist?(base, '.mytaxa') ||
         result_files_exist?(base, '.nomytaxa.txt')
-      r = MiGA::Result.new("#{base}.json")
       add_files_to_ds_result(
-        r, name,
+        MiGA::Result.new("#{base}.json"), name,
         mytaxa: '.mytaxa',
         blast: '.blast',
         mytaxain: '.mytaxain',
@@ -344,7 +347,8 @@ module MiGA::Dataset::Result
         phylum: '.mytaxa.Phylum.txt',
         innominate: '.mytaxa.innominate',
         kronain: '.mytaxa.krona',
-        krona: '.html')
+        krona: '.html'
+      )
     else
       MiGA::Result.new("#{base}.json")
     end
@@ -357,9 +361,8 @@ module MiGA::Dataset::Result
       return nil unless
         result_files_exist?(base, %w[.pdf .mytaxa]) ||
         result_files_exist?(base, '.nomytaxa.txt')
-      r = MiGA::Result.new("#{base}.json")
       add_files_to_ds_result(
-        r, name,
+        MiGA::Result.new("#{base}.json"), name,
         nomytaxa: '.nomytaxa.txt',
         mytaxa: '.mytaxa',
         report: '.pdf',
@@ -370,7 +373,8 @@ module MiGA::Dataset::Result
         wintax: '.wintax',
         gene_ids: '.wintax.genes',
         region_ids: '.wintax.regions',
-        regions: '.reg')
+        regions: '.reg'
+      )
     else
       MiGA::Result.new("#{base}.json")
     end
@@ -428,9 +432,8 @@ module MiGA::Dataset::Result
     return nil unless
       result_files_exist?(base, %w[.aai-medoids.tsv .aai.db]) ||
       result_files_exist?(base, %w[.ani-medoids.tsv .ani.db])
-    r = MiGA::Result.new("#{base}.json")
     add_files_to_ds_result(
-      r, name,
+      MiGA::Result.new("#{base}.json"), name,
       aai_medoids: '.aai-medoids.tsv',
       haai_db: '.haai.db',
       aai_db: '.aai.db',
@@ -438,7 +441,8 @@ module MiGA::Dataset::Result
       ani_db: '.ani.db',
       ref_tree: '.nwk',
       ref_tree_pdf: '.nwk.pdf',
-      intax_test: '.intax.txt')
+      intax_test: '.intax.txt'
+    )
   end
 
   ##

data/lib/miga/metadata.rb

@@ -27,7 +27,7 @@ class MiGA::Metadata < MiGA::MiGA
 
   ##
   # Initiate a MiGA::Metadata object with description in +path+.
-  # It will create it if it doesn't exist
+  # It will create it if it doesn't exist.
   def initialize(path, defaults = {})
     @data = nil
     @path = File.absolute_path(path)
@@ -39,21 +39,21 @@ class MiGA::Metadata < MiGA::MiGA
   end
   
   ##
-  # Parsed data as a Hash.
+  # Parsed data as a Hash
   def data
     self.load if @data.nil?
     @data
   end
 
   ##
-  # Reset :created field and save the current data.
+  # Reset :created field and save the current data
   def create
     self[:created] = Time.now.to_s
     save
   end
 
   ##
-  # Save the metadata into #path.
+  # Save the metadata into #path
   def save
     MiGA.DEBUG "Metadata.save #{path}"
     self[:updated] = Time.now.to_s
@@ -78,7 +78,7 @@ class MiGA::Metadata < MiGA::MiGA
   end
 
   ##
-  # (Re-)load metadata stored in #path.
+  # (Re-)load metadata stored in #path
   def load
     sleeper = 0.0
     while File.exist? lock_file
@@ -87,11 +87,11 @@ class MiGA::Metadata < MiGA::MiGA
     end
     tmp = MiGA::Json.parse(path, additions: true)
     @data = {}
-    tmp.each_pair{ |k,v| self[k] = v }
+    tmp.each { |k, v| self[k] = v }
   end
 
   ##
-  # Delete file at #path.
+  # Delete file at #path
   def remove!
     MiGA.DEBUG "Metadata.remove! #{path}"
     File.unlink(path)
@@ -99,29 +99,41 @@ class MiGA::Metadata < MiGA::MiGA
   end
 
   ##
-  # Lock file for the metadata.
-  def lock_file ; "#{path}.lock" ; end
+  # Lock file for the metadata
+  def lock_file
+    "#{path}.lock"
+  end
 
   ##
-  # Return the value of +k+ in #data.
-  def [](k) data[k.to_sym] end
+  # Return the value of +k+ in #data
+  def [](k)
+    data[k.to_sym]
+  end
 
   ##
-  # Set the value of +k+ to +v+.
-  def []=(k,v)
+  # Set the value of +k+ to +v+
+  def []=(k, v)
     self.load if @data.nil?
     k = k.to_sym
-    # Protect the special field :name
-    v=v.miga_name if k==:name
-    # Symbolize the special field :type
-    v=v.to_sym if k==:type
-    # Delete if nil, register, and return
-    v.nil? ? @data.delete(k) : (@data[k]=v)
+    return @data.delete(k) if v.nil?
+
+    case k
+    when :name
+      # Protect the special field :name
+      v = v.miga_name
+    when :type
+      # Symbolize the special field :type
+      v = v.to_sym if k == :type
+    end
+
+    @data[k] = v
   end
 
   ##
-  # Iterate +blk+ for each data with 2 arguments key and value.
-  def each(&blk) data.each{ |k,v| blk.call(k,v) } ; end
+  # Iterate +blk+ for each data with 2 arguments: key and value
+  def each(&blk)
+    data.each { |k, v| blk.call(k, v) }
+  end
 
   ##
   # Show contents in JSON format as a String

data/lib/miga/result/dates.rb

@@ -30,16 +30,16 @@ module MiGA::Result::Dates
 
   private
     
-    ##
-    # Internal function to detect start and end dates
-    def date_at(event)
-      date = self[event]
-      date ||= self[:started] if event == :start
-      if date.nil?
-        f = path event
-        date = File.read(f) if File.size? f
-      end
-      date.nil? ? nil : Time.parse(date)
+  ##
+  # Internal function to detect start and end dates
+  def date_at(event)
+    date = self[event]
+    date ||= self[:started] if event == :start
+    if date.nil?
+      f = path event
+      date = File.read(f) if File.size? f
     end
+    Time.parse(date) unless date.nil?
+  end
 end
 

data/lib/miga/tax_dist.rb

@@ -1,66 +1,82 @@
 # @package MiGA
 # @license Artistic-2.0
 
-require "miga/common"
-require "miga/taxonomy"
-require "zlib"
+require 'miga/common'
+require 'miga/taxonomy'
+require 'zlib'
 
 ##
 # Methods for taxonomy identification based on AAI/ANI values.
 module MiGA::TaxDist
-  
-  ##
-  # Absolute path to the :intax or :novel data file (determined by +test+) for
-  # AAI.
-  def self.aai_path(test)
-    test = test.downcase.to_sym
-    return nil unless [:intax, :novel].include? test
-    File.expand_path("../_data/aai-#{test}.tsv.gz", __FILE__)
-  end
+  # Class-level
+  class << self
+
+    ##
+    # Absolute path to the :intax or :novel data file (determined by +test+) for
+    # AAI, determined for options +opts+. Supported options:
+    # - +:engine+: The search engine for AAI: +:blast+ (default) or +:diamond+
+    def aai_path(test, opts = {})
+      opts[:engine] ||= :blast
+      engine = opts[:engine].to_s.downcase.to_sym
+      test = test.to_s.downcase.to_sym
+      return nil unless %i[intax novel].include? test
+      engine = :blast if %i[blast+ blat].include? engine
+      return nil unless %i[blast diamond].include? engine
+      File.expand_path("../_data/aai-#{test}-#{engine}.tsv.gz", __FILE__)
+    end
 
-  # Returns a Hash, where the keys correspond to the taxonomic level
-  # (see MiGA::Taxonomy.LONG_RANKS for the meanings), and the values correspond
-  # to the p-values of being :intax or :novel, as determined by +test+.
-  def self.aai_pvalues(aai, test)
-    Zlib::GzipReader.open(aai_path(test)) do |fh|
-      keys = nil
-      fh.each_line do |ln|
-        row = ln.chomp.split(/\t/)
-        if fh.lineno==1
-          keys = row[1, row.size-1].map{ |i| i.to_i }
-        elsif row.shift.to_f >= aai
-          vals = {}
-          keys.each do |i|
-            v = row.shift
-            next if v=="NA"
-            rank = i==0 ? :root : MiGA::Taxonomy.KNOWN_RANKS[i-1]
-            vals[rank] = v.to_f
+    ##
+    # Returns a Hash, where the keys correspond to the taxonomic level
+    # (see MiGA::Taxonomy.LONG_RANKS for the meanings), and the values
+    # correspond to the p-values of +test+ (one of +:intax+ or +:novel+)
+    # with options +opts+. See +aai_path+ for supported options.
+    def aai_pvalues(aai, test, opts = {})
+      Zlib::GzipReader.open(aai_path(test, opts)) do |fh|
+        keys = nil
+        fh.each_line do |ln|
+          row = ln.chomp.split(/\t/)
+          if fh.lineno == 1
+            keys = row[1, row.size - 1].map(&:to_i)
+          elsif row.shift.to_f >= aai
+            vals = {}
+            keys.each do |i|
+              v = row.shift
+              next if v == 'NA' # <- missing data
+              next if i == 1 # <- namespace, not a taxonomic rank
+              rank = i.zero? ? :root : MiGA::Taxonomy.KNOWN_RANKS[i]
+              vals[rank] = v.to_f
+            end
+            return vals
           end
-          return vals
         end
-      end # each_line ln
-    end # open fh
-    {}
-  end
+      end
+      {}
+    end
 
-  # Determines the degree to which a Float +aai+ value indicates similar
-  # taxonomy (with +test+ :intax) or a novel taxon (with +test+ :novel). Returns
-  # a Hash with "likelihood" phrases as keys and values as an array with
-  # cannonical rank (as in MiGA::Taxonomy) and estimated p-value.
-  def self.aai_taxtest(aai, test)
-    meaning = {most_likely:[0,0.01],probably:[0.01,0.1],possibly_even:[0.1,0.5]}
-    pv = aai_pvalues(aai, test)
-    out = {}
-    meaning.each do |phrase, thresholds|
-      lwr, upr = thresholds
-      min = pv.values.select{ |v| v < upr }.max
-      return out if min.nil?
-      if min >= lwr
-        v = pv.select{ |_,vj| vj==min }
-        out[phrase] = (test==:intax ? v.reverse_each : v).first
+    ##
+    # Determines the degree to which a Float +aai+ value indicates similar
+    # taxonomy (with +test+ :intax) or a novel taxon (with +test+ :novel) with
+    # options +opts+. See +aai_path+ for supported options.
+    # Returns a Hash with "likelihood" phrases as keys and values as an array
+    # with cannonical rank (as in MiGA::Taxonomy) and estimated p-value.
+    def aai_taxtest(aai, test, opts = {})
+      meaning = {
+        most_likely:   [0.00, 0.01],
+        probably:      [0.01, 0.10],
+        possibly_even: [0.10, 0.50]
+      }
+      pvalues = aai_pvalues(aai, test, opts)
+      out = {}
+      meaning.each do |phrase, thresholds|
+        lwr, upr = thresholds
+        min = pvalues.values.select { |v| v < upr }.max
+        return out if min.nil?
+        if min >= lwr
+          v = pvalues.select { |_, vj| vj == min }
+          out[phrase] = (test == :intax ? v.reverse_each : v).first
+        end
       end
+      out
     end
-    out
   end
-
 end

data/lib/miga/taxonomy.rb

@@ -35,7 +35,6 @@ class MiGA::Taxonomy < MiGA::MiGA
     else
       initialize_by_ranks(str, ranks)
     end
-    initialize_by_str(str)
   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 = [0.6, 2, 0]
+  VERSION = [0.6, 3, 0]
 
   ##
   # Nickname for the current major.minor version.
@@ -18,7 +18,7 @@ module MiGA
 
   ##
   # Date of the current gem release.
-  VERSION_DATE = Date.new(2020, 3, 11)
+  VERSION_DATE = Date.new(2020, 3, 27)
 
   ##
   # Reference of MiGA.

data/test/result_stats_test.rb

@@ -0,0 +1,119 @@
+require 'test_helper'
+require 'miga/project'
+
+class ResultStatsTest < Test::Unit::TestCase
+
+  def setup
+    $tmp = Dir.mktmpdir
+    ENV['MIGA_HOME'] = $tmp
+    FileUtils.touch(File.expand_path('.miga_rc', ENV['MIGA_HOME']))
+    FileUtils.touch(File.expand_path('.miga_daemon.json', ENV['MIGA_HOME']))
+    $p = MiGA::Project.new(File.expand_path('project1', $tmp))
+    $d = $p.add_dataset('dataset1')
+  end
+
+  def teardown
+    FileUtils.rm_rf $tmp
+    ENV['MIGA_HOME'] = nil
+  end
+
+  def file_path(dir, ext)
+    File.join($p.path, dir, "#{$d.name}#{ext}")
+  end
+
+  def touch_done(dir)
+    FileUtils.touch(file_path(dir, '.done'))
+  end
+
+  def test_single_raw_reads
+    dir = 'data/01.raw_reads'
+    fq = file_path(dir, '.1.fastq')
+    File.open(fq, 'w') { |fh| fh.puts '@1','ACTAC','+','####' }
+    touch_done(dir)
+    r = $d.add_result(:raw_reads)
+    assert_equal({}, r[:stats])
+    r.compute_stats
+    assert(!r[:stats].empty?)
+    assert_equal(Hash, r[:stats].class)
+    assert_equal(1, r[:stats][:reads])
+    assert_equal([40.0, '%'], r[:stats][:g_c_content])
+  end
+
+  def test_coupled_raw_reads
+    dir = 'data/01.raw_reads'
+    fq = file_path(dir, '.1.fastq')
+    File.open(fq, 'w') { |fh| fh.puts '@1','ACTAC','+','####' }
+    fq = file_path(dir, '.2.fastq')
+    File.open(fq, 'w') { |fh| fh.puts '@1','ACTAC','+','####' }
+    touch_done(dir)
+    r = $d.add_result(:raw_reads)
+    r.compute_stats
+    assert(!r[:stats].empty?)
+    assert_nil(r[:stats][:reads])
+    assert_equal(1, r[:stats][:read_pairs])
+    assert_equal([40.0, '%'], r[:stats][:reverse_g_c_content])
+  end
+
+  def test_trimmed_reads
+    dir = 'data/02.trimmed_reads'
+    FileUtils.touch(file_path(dir, '.1.clipped.fastq'))
+    touch_done(dir)
+    r = $d.add_result(:trimmed_reads)
+    assert_equal({}, r[:stats])
+    r.compute_stats
+    assert_equal({}, r[:stats])
+  end
+
+  def test_read_quality
+    dir = 'data/03.read_quality'
+    Dir.mkdir(file_path(dir, '.solexaqa'))
+    Dir.mkdir(file_path(dir, '.fastqc'))
+    touch_done(dir)
+    r = $d.add_result(:read_quality)
+    assert_equal({}, r[:stats])
+    r.compute_stats
+    assert_equal({}, r[:stats])
+  end
+
+  def test_trimmed_fasta
+    dir = 'data/04.trimmed_fasta'
+    fa = file_path(dir, '.CoupledReads.fa')
+    File.open(fa, 'w') { |fh| fh.puts '>1','ACTAC' }
+    touch_done(dir)
+    r = $d.add_result(:trimmed_fasta)
+    assert_equal({}, r[:stats])
+    r.compute_stats
+    assert_equal(1, r[:stats][:reads])
+    assert_equal([40.0, '%'], r[:stats][:g_c_content])
+  end
+
+  def test_assembly
+    dir = 'data/05.assembly'
+    fa = file_path(dir, '.LargeContigs.fna')
+    File.open(fa, 'w') { |fh| fh.puts '>1','ACTAC' }
+    touch_done(dir)
+    r = $d.add_result(:assembly)
+    assert_equal({}, r[:stats])
+    r.compute_stats
+    assert_equal(1, r[:stats][:contigs])
+    assert_equal([5, 'bp'], r[:stats][:total_length])
+    assert_equal([40.0, '%'], r[:stats][:g_c_content])
+  end
+
+  def test_cds
+    dir = 'data/06.cds'
+    fa = file_path(dir, '.faa')
+    File.open(fa, 'w') { |fh| fh.puts '>1','M' }
+    touch_done(dir)
+    r = $d.add_result(:cds)
+    assert_equal({}, r[:stats])
+    r.compute_stats
+    assert_equal(1, r[:stats][:predicted_proteins])
+    assert_equal([1.0, 'aa'], r[:stats][:average_length])
+    assert_nil(r[:stats][:coding_density])
+    test_assembly
+    r.compute_stats
+    assert_equal([60.0, '%'], r[:stats][:coding_density])
+  end
+
+end

data/test/result_test.rb

@@ -10,14 +10,14 @@ class ResultTest < Test::Unit::TestCase
     FileUtils.touch(File.expand_path('.miga_daemon.json', ENV['MIGA_HOME']))
     $p1 = MiGA::Project.new(File.expand_path('project1', $tmp))
     $d1 = $p1.add_dataset('dataset1')
-    FileUtils.touch(File.expand_path(
-      "data/02.trimmed_reads/#{$d1.name}.1.clipped.fastq", $p1.path))
-    FileUtils.touch(File.expand_path(
-      "data/02.trimmed_reads/#{$d1.name}.done", $p1.path))
-    FileUtils.touch(File.expand_path(
-      'data/10.clades/01.find/miga-project.empty', $p1.path))
-    FileUtils.touch(File.expand_path(
-      'data/10.clades/01.find/miga-project.done', $p1.path))
+    FileUtils.touch(
+      File.join($p1.path, "data/02.trimmed_reads/#{$d1.name}.1.clipped.fastq"))
+    FileUtils.touch(
+      File.join($p1.path, "data/02.trimmed_reads/#{$d1.name}.done"))
+    FileUtils.touch(
+      File.join($p1.path, 'data/10.clades/01.find/miga-project.empty'))
+    FileUtils.touch(
+      File.join($p1.path, 'data/10.clades/01.find/miga-project.done'))
   end
 
   def teardown
@@ -46,4 +46,18 @@ class ResultTest < Test::Unit::TestCase
     assert_equal($p1.path, r.source.path)
   end
 
+  def test_dates
+    r = $d1.add_result(:trimmed_reads)
+    assert_nil(r.done_at)
+    assert_nil(r.started_at)
+    tf = File.join($p1.path, "data/02.trimmed_reads/#{$d1.name}.done")
+    File.open(tf, 'w') { |fh| fh.puts Time.new(1,2,3,4,5) }
+    assert_equal(Time, r.done_at.class)
+    assert_nil(r.running_time)
+    tf = File.join($p1.path, "data/02.trimmed_reads/#{$d1.name}.start")
+    File.open(tf, 'w') { |fh| fh.puts Time.new(1,2,3,4,0) }
+    r = $d1.add_result(:trimmed_reads)
+    assert_equal(5.0, r.running_time)
+  end
+
 end

data/test/tax_dist_test.rb

@@ -0,0 +1,59 @@
+require 'test_helper'
+require 'miga/tax_dist'
+
+class TaxDistTest < Test::Unit::TestCase
+
+  def test_aai_path
+    assert(File.size? MiGA::TaxDist.aai_path(:intax))
+    assert(File.size? MiGA::TaxDist.aai_path(:novel))
+    assert(File.size? MiGA::TaxDist.aai_path(:intax, engine: :diamond))
+    assert(File.size? MiGA::TaxDist.aai_path(:novel, engine: :blast))
+    assert(File.size? MiGA::TaxDist.aai_path(:novel, engine: :'blast+'))
+  end
+
+  def test_aai_pvalues
+    distant_intax = MiGA::TaxDist.aai_pvalues(35.0, :intax)
+    assert_lt(distant_intax[:root], 0.05)
+    assert_gt(distant_intax[:g], 0.05)
+    assert_nil(distant_intax[:ns])
+
+    close_intax = MiGA::TaxDist.aai_pvalues(99.0, :intax, engine: :blast)
+    assert_lt(close_intax[:root], 0.05)
+    assert_lt(close_intax[:s], 0.05)
+
+    close_intax = MiGA::TaxDist.aai_pvalues(99.0, :intax, engine: :diamond)
+    assert_lt(close_intax[:root], 0.05)
+    assert_lt(close_intax[:s], 0.05)
+
+    distant_novel = MiGA::TaxDist.aai_pvalues(35.0, :novel, engine: :diamond)
+    $stderr.puts distant_novel
+    assert_gt(distant_novel[:root], 0.05)
+    assert_lt(distant_novel[:g], 0.05)
+    assert_nil(distant_novel[:ns])
+
+    close_novel = MiGA::TaxDist.aai_pvalues(99.0, :novel)
+    assert_gt(close_novel[:root], 0.05)
+    assert_gt(close_novel[:f], 0.05)
+
+    assert_equal({}, MiGA::TaxDist.aai_pvalues(101.0, :intax))
+  end
+
+  def test_aai_taxtest
+    distant_intax = MiGA::TaxDist.aai_taxtest(35.0, :intax, engine: :diamond)
+    assert_equal(:root, distant_intax[:most_likely][0])
+    assert_nil(distant_intax[:probably])
+    assert_nil(distant_intax[:possibly_even])
+
+    distant_intax = MiGA::TaxDist.aai_taxtest(35.0, :intax, engine: :blast)
+    assert_equal(:root, distant_intax[:most_likely][0])
+    assert_nil(distant_intax[:probably])
+    assert_nil(distant_intax[:possibly_even])
+
+    close_intax = MiGA::TaxDist.aai_taxtest(99.0, :intax, engine: :diamond)
+    assert_equal(:s, close_intax[:probably][0])
+
+    close_intax = MiGA::TaxDist.aai_taxtest(99.0, :intax, engine: :blast)
+    assert_equal(:s, close_intax[:probably][0])
+  end
+
+end

data/test/test_helper.rb

@@ -3,6 +3,7 @@ SimpleCov.start
 
 require 'rubygems'
 require 'test/unit'
+require 'assertions'
 require 'miga/common'
 require 'stringio'
 

data/utils/distance/commands.rb

@@ -63,12 +63,12 @@ module MiGA::DistanceRunner::Commands
 
   ##
   # Execute an AAI command
-  def aai_cmd(f1, f2, n1, n2, db, o={})
+  def aai_cmd(f1, f2, n1, n2, db, o = {})
     o = opts.merge(o)
     v = `aai.rb -1 "#{f1}" -2 "#{f2}" -S "#{db}" \
           --name1 "#{n1}" --name2 "#{n2}" \
           -t "#{o[:thr]}" -a --lookup-first "--#{o[:aai_save_rbm]}" \
-          -p "#{o[:aai_p] || "blast+"}"`.chomp
+          -p "#{o[:aai_p] || 'blast+'}"`.chomp
     (v.nil? || v.empty?) ? 0 : v.to_f
   end
 
@@ -91,7 +91,7 @@ module MiGA::DistanceRunner::Commands
       v = `ani.rb -1 "#{f1}" -2 "#{f2}" -S "#{db}" \
             --name1 "#{n1}" --name2 "#{n2}" \
             -t "#{opts[:thr]}" -a --no-save-regions --no-save-rbm \
-            --lookup-first -p "#{o[:ani_p] || "blast+"}"`.chomp
+            --lookup-first -p "#{o[:ani_p] || 'blast+'}"`.chomp
     end
     v.nil? || v.empty? ? 0 : v.to_f
   end

data/utils/distance/pipeline.rb

@@ -73,14 +73,17 @@ module MiGA::DistanceRunner::Pipeline
     cr = dataset.closest_relatives(1, from_ref_project)
     return if cr.nil? or cr.empty?
     tax = ref_project.dataset(cr[0][0]).metadata[:tax] || {}
+
     # Run the test for each rank
-    r = MiGA::TaxDist.aai_pvalues(cr[0][1], :intax).map do |k,v|
+    tax_test = MiGA::TaxDist.aai_pvalues(cr[0][1], :intax, engine: opts[:aai_p])
+    r = tax_test.map do |k,v|
       sig = ''
-      [0.5,0.1,0.05,0.01].each{ |i| sig << '*' if v<i }
+      [0.5, 0.1, 0.05, 0.01].each { |i| sig << '*' if v < i }
       [MiGA::Taxonomy.LONG_RANKS[k], (tax[k] || '?'), v, sig]
     end
+
     # Save test
-    File.open(File.expand_path("#{dataset.name}.intax.txt", home), "w") do |fh|
+    File.open(File.expand_path("#{dataset.name}.intax.txt", home), 'w') do |fh|
       fh.puts "Closest relative: #{cr[0][0]} with AAI: #{cr[0][1]}."
       fh.puts ''
       fh.puts MiGA::MiGA.tabulate(%w[Rank Taxonomy P-value Signif.], r)

data/utils/distance/runner.rb

@@ -22,6 +22,7 @@ class MiGA::DistanceRunner
       raise "No project at #{project_path}"
     @dataset = project.dataset(dataset_name)
     @home = File.expand_path('data/09.distances', project.path)
+
     # Default opts
     if project.metadata[:aai_save_rbm] == false
       @opts[:aai_save_rbm] ||= 'no-save-rbm'
@@ -48,6 +49,8 @@ class MiGA::DistanceRunner
     [:haai_p, :aai_p, :ani_p, :distances_checkpoint].each do |m|
       @opts[m] ||= ref_project.metadata[m]
     end
+    @opts[:aai_p] ||= 'blast+'
+    @opts[:ani_p] ||= 'blast+'
     @opts[:distances_checkpoint] ||= 10
     @opts[:distances_checkpoint] = @opts[:distances_checkpoint].to_i
   end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: miga-base
 version: !ruby/object:Gem::Version
-  version: 0.6.2.0
+  version: 0.6.3.0
 platform: ruby
 authors:
 - Luis M. Rodriguez-R
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-03-11 00:00:00.000000000 Z
+date: 2020-03-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: daemons
@@ -80,6 +80,20 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '3'
+- !ruby/object:Gem::Dependency
+  name: assertions
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '1'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '1'
 description: Microbial Genomes Atlas
 email: lmrodriguezr@gmail.com
 executables:
@@ -94,8 +108,10 @@ files:
 - Rakefile
 - bin/miga
 - lib/miga.rb
-- lib/miga/_data/aai-intax.tsv.gz
-- lib/miga/_data/aai-novel.tsv.gz
+- lib/miga/_data/aai-intax-blast.tsv.gz
+- lib/miga/_data/aai-intax-diamond.tsv.gz
+- lib/miga/_data/aai-novel-blast.tsv.gz
+- lib/miga/_data/aai-novel-diamond.tsv.gz
 - lib/miga/cli.rb
 - lib/miga/cli/action.rb
 - lib/miga/cli/action/about.rb
@@ -199,7 +215,9 @@ files:
 - test/metadata_test.rb
 - test/project_test.rb
 - test/remote_dataset_test.rb
+- test/result_stats_test.rb
 - test/result_test.rb
+- test/tax_dist_test.rb
 - test/tax_index_test.rb
 - test/taxonomy_test.rb
 - test/test_helper.rb