module MiGA::TaxDist

Methods for taxonomy identification based on AAI/ANI values.

Public Class Methods

aai_path(test, opts = {}) click to toggle source

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

# File lib/miga/tax_dist.rb, line 18
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 %[intax novel].include? test
  engine = :blast if %[blast+ blat].include? engine
  return nil unless %[blast diamond].include? engine
  File.expand_path("../_data/aai-#{test}-#{engine}.tsv.gz", __FILE__)
end
aai_pvalues(aai, test, opts = {}) click to toggle source

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.

# File lib/miga/tax_dist.rb, line 33
def aai_pvalues(aai, test, opts = {})
  y = {}
  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
        y = vals
        break
      end
    end
    fh.rewind # to avoid warnings caused by the break above
  end
  y
end
aai_taxtest(aai, test, opts = {}) click to toggle source

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.

# File lib/miga/tax_dist.rb, line 65
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