Module: Annotations::Genes

Defined in:

lib/MARQ/annotations.rb

Defined Under Namespace

Modules: Genecodis, SENT

Class Method Summary

• .get_genes(dataset, options = {}) ⇒ Object
• .get_genes_nth(dataset, num_genes) ⇒ Object
• .get_genes_old(dataset, cut_off = 0.1, fdr = false) ⇒ Object

Class Method Details

.get_genes(dataset, options = {}) ⇒ Object

# File 'lib/MARQ/annotations.rb', line 538

def self.get_genes(dataset, options = {})
  fdr, cut_off, folds, do_folds, nth_genes = {
    :fdr => false,
    :cut_off => 0.05,
    :folds => 2.5,
    :do_folds => true,
    :nth_genes => 0,
  }.merge(options).values_at(:fdr, :cut_off, :folds, :do_folds, :nth_genes)

  if nth_genes > 0
    return get_genes_nth(dataset, nth_genes)
  end


  path = MARQ.dataset_path(dataset)

  experiments = File.open(path + '.experiments').collect{|l| l.chomp.strip}
  genes       = File.open(path + '.codes').collect{|l| l.chomp.strip}


  experiments_ts   = experiments.select{|exp| exp !~ /\[ratio\]/}
  experiments_fold = experiments.select{|exp| exp =~ /\[ratio\]/}

  experiments_fold = [] if ! do_folds

  values_up = {}
  values_down = {}
  experiments.each{|exp| values_up[exp] = []; values_down[exp] = []}

  File.open(path + '.pvalues').each_with_index{|l, i|
    values = l.chomp.split(/\t/)
    experiments_ts.zip(values).each{|p|
      name = p.first
      value = p.last == "NA" ? 1.0 : p.last.to_f
      values_up[name]     << (value > 0 ? value : 1.0)
      values_down[name]   << (value < 0 ? - value : 1.0)
    }
  }

  File.open(path + '.logratios').each_with_index{|l, i|
    values = l.chomp.split(/\t/)
    experiments.zip(values).each{|p|
      name = p.first
      next unless experiments_fold.include? name
      value = p.last == "NA" ? 0 : p.last.to_f
      values_up[name]     << (p.last > 0 ? value : 0)
      values_down[name]   << (p.last < 0 ? - value : 0)
    }
  }

  genes_up = {}
  genes_down = {}

  threshold = cut_off
  values_up.each{|experiment, values|
    genes_up[experiment] = []
    if experiments_ts.include? experiment
      if fdr
        threshold = FDR.step_up(values.sort, cut_off)
        next if threshold == 0.0
      end
      values.each_with_index{|value, i| genes_up[experiment] << genes[i] if value < threshold}
    elsif experiment_fold.include? experiment
      values.each_with_index{|value, i| genes_up[experiment] << genes[i] if value < folds}
    end
  }
  values_down.each{|experiment, values|
    genes_down[experiment] = []
    if experiments_ts.include? experiment
      if fdr
        threshold = FDR.step_up(values.sort, cut_off)
        next if threshold == 0.0
      end
      values.each_with_index{|value, i| genes_down[experiment] << genes[i] if value < threshold}
    elsif experiment_fold.include? experiment
      values.each_with_index{|value, i| genes_down[experiment] << genes[i] if value < folds}
    end
  }


  {:up => genes_up, :down => genes_down}
end

.get_genes_nth(dataset, num_genes) ⇒ Object

# File 'lib/MARQ/annotations.rb', line 513

def self.get_genes_nth(dataset, num_genes)
  path = MARQ::Dataset.path(dataset)

  experiments = File.open(path + '.experiments').collect{|l| l.chomp.strip}
  genes       = File.open(path + '.codes').collect{|l| l.chomp.strip}
  total_genes = genes.length

  genes_up = {}
  genes_down = {}
  experiments.each{|exp| genes_up[exp] = []; genes_down[exp] = []}

  File.open(path + '.orders').each_with_index{|l, i|
    values = l.chomp.split(/\t/)
    experiments.zip(values).each{|p|
      name = p.first
      value = p.last
      next if p.last == "NA"
      genes_up[name]   <<  genes[i] if value.to_i < num_genes
      genes_down[name] <<  genes[i] if value.to_i > total_genes - num_genes
    }
  }

  {:up => genes_up, :down => genes_down}
end

.get_genes_old(dataset, cut_off = 0.1, fdr = false) ⇒ Object

# File 'lib/MARQ/annotations.rb', line 621

def self.get_genes_old(dataset, cut_off = 0.1, fdr = false)

  path = MARQ::Dataset.path(dataset)

  experiments = File.open(path + '.experiments').collect{|l| l.chomp.strip}.select{|name| !name.match(/\[ratio\]/)}
  genes = File.open(path + '.codes').collect{|l| l.chomp.strip}


  values_up = {}
  values_down = {}
  experiments.each{|exp| values_up[exp] = []; values_down[exp] = []}


  if File.exist?(path + '.pvalues')
    File.open(path + '.pvalues').each_with_index{|l, i|
      values = l.chomp.split(/\t/)
      experiments.zip(values).each{|p|
        value = p.last == "NA" ? 1.0 : p.last.to_f
        values_up[p.first]     << (value > 0 ? value : 1.0)
        values_down[p.first]   << (value < 0 ? - value : 1.0)
      }
    }
  end

  genes_up = {}
  genes_down = {}

  threshold = cut_off
  values_up.each{|experiment, values|
    genes_up[experiment] = []
    if fdr
      threshold = FDR.step_up(values.sort, cut_off)
      next if threshold == 0.0
    end
    values.each_with_index{|value, i| genes_up[experiment] << genes[i] if value < threshold}
  }
  values_down.each{|experiment, values|
    genes_down[experiment] = []
    if fdr
      threshold = FDR.step_up(values.sort, cut_off)
      next if threshold == 0.0
    end
    values.each_with_index{|value, i| genes_down[experiment] << genes[i] if value < threshold}
  }


  {:up => genes_up, :down => genes_down}
end