module MiGA::Cli::Action::Doctor::Base

check_dist_eval(cli, p, res) click to toggle source

Scans the all-vs-all matrix registered in res (MiGA::Result) in search of pairs where one or both datasets are missing or inactive in the project p (MiGA::Project), and report progress through cli (MiGA::Cli). Returns an Array with two arrays: the first a list of dataset names that are no longer registered in the project or are currently inactive, and the second a list of dataset names that have registered pairs with the first list, and therefore the databases need to be cleaned. This is a subtask of check_dist

# File lib/miga/cli/action/doctor/base.rb, line 14
def check_dist_eval(cli, p, res)
  notok = {}
  fix = {}
  Zlib::GzipReader.open(res.file_path(:matrix)) do |fh|
    lineno = 0
    fh.each_line do |ln|
      next if (lineno += 1) == 1

      r = ln.split("\t")
      names = [r[0], r[1]]
      next unless names.any? { |i| p.dataset(i).nil? }

      names.each do |i|
        if p.dataset(i).nil? || !p.dataset(i).active?
          notok[i] = true
        else
          fix[i] = true
        end
      end
    end
  end
  [notok.keys, fix.keys]
end

check_dist_fix(cli, p, fix) click to toggle source

Cleanup distance databases for datasets names in fix (Array: String) from project p (MiGA::Project), and report through cli (MiGA::Cli). This is a subtask of check_dist

# File lib/miga/cli/action/doctor/base.rb, line 42
def check_dist_fix(cli, p, fix)
  return if fix.empty?

  cli.say("- Fixing #{fix.size} datasets")
  fix.each do |d_n|
    cli.say "  > Fixing #{d_n}."
    p.dataset(d_n).cleanup_distances!
  end
end

check_dist_recompute(cli, res, notok) click to toggle source

Recompute res (MiGA::Result) if notok (Array: String) has any dataset names registered, and report through cli (MiGA::Cli). This is a subtask of check_dist

# File lib/miga/cli/action/doctor/base.rb, line 56
def check_dist_recompute(cli, res, notok)
  return if notok.empty?

  cli.say '- Unregistered datasets detected: '
  if notok.size <= 5
    notok.each { |i| cli.say "  > #{i}" }
  else
    cli.say "  > #{notok.size}, including #{notok.first}"
  end
  cli.say '- Removing tables, recompute'
  res.remove!
end

outdated_fastaai_ess(res) click to toggle source

Check if the essential genes result res has an outdated FastAAI index

# File lib/miga/cli/action/doctor/base.rb, line 131
def outdated_fastaai_ess(res)
  idx1 = res.file_path(:fastaai_index)
  idx2 = res.file_path(:fastaai_index_2)
  idx3 = res.file_path(:fastaai_crystal)
  idx3.nil? && !(idx1.nil? && idx2.nil?)
end

read_bidirectional(a, metric) click to toggle source

Reads all the distance estimates in a -> * for metric and returns them as a hash +{“b_name” => [val, sd, …], …}+

# File lib/miga/cli/action/doctor/base.rb, line 82
def read_bidirectional(a, metric)
  db_file = a.result(:distances)&.file_path("#{metric}_db") or return {}
  sql = "select seq2, #{metric}, sd, n, omega from #{metric}"
  data = MiGA::SQLite.new(db_file).run(sql) || []
  Hash[
    data.map do |row|
      k, v = row.shift(2)
      [k, row.all?(&:zero?) ? v : [v] + row]
    end
  ]
end

run_cmd(cmd, opts = {}) click to toggle source

Run command cmd with options opts

# File lib/miga/cli/action/doctor/base.rb, line 123
def run_cmd(cmd, opts = {})
  opts = { return: :output, err2out: true, raise: false }.merge(opts)
  cmdo = MiGA::MiGA.run_cmd(cmd, opts).chomp
  warn(cmdo) unless cmdo.empty?
end

save_bidirectional(a, dist) click to toggle source

Saves all the distance estimates in * -> a into the a databases (as a -> *), where a is a MiGA::Dataset object, with currently saved values read from the hash dist

# File lib/miga/cli/action/doctor/base.rb, line 98
def save_bidirectional(a, dist)
  each_database_file(a) do |db_file, metric, result, rank|
    next if rank == :haai # No need for hAAI to be bidirectional
    next if result == :taxonomy # Taxonomy is never bidirectional

    b2a = dist[rank].map { |b_name, v| b_name if v[a.name] }.compact
    a2b = dist[rank][a.name]&.keys || []
    MiGA::SQLite.new(db_file).run do |db|
      sql = <<~SQL
        insert into #{metric}(seq1, seq2, #{metric}, sd, n, omega) \
        values(, ?, , ?, , ?);
      SQL
      db.execute('BEGIN TRANSACTION;')
      (b2a - a2b).each do |b_name|
        val = dist[rank][b_name][a.name]
        val = [val, 0, 0, 0] unless val.is_a?(Array)
        db.execute(sql, [a.name, b_name] + val)
      end
      db.execute('COMMIT;')
    end
  end
end

saved_targets(dataset) click to toggle source

Returns all targets identified by AAI

# File lib/miga/cli/action/doctor/base.rb, line 71
def saved_targets(dataset)
  # Return nil if distance or database are not retrievable
  dist = dataset.result(:distances) or return
  path = dist.file_path(:aai_db) or return

  MiGA::SQLite.new(path).run('select seq2 from aai').map(&:first)
end

module MiGA::Cli::Action::Doctor::Base

Public Instance Methods