Sha256: e299f7216f0670f31d9a696dfe16864d5cc3d2fd9bb501d90e6f1306f9988da3
Contents?: true
Size: 1.56 KB
Versions: 11
Compression:
Stored size: 1.56 KB
Contents
##
# FILTER UC FILE BY MIN SAMPLES
#
module Lederhosen
class CLI
desc "uc_filter",
"filter UCLUST output to remove small, infrequent clusters"
method_option :input, :type => :string, :required => true
method_option :output, :type => :string, :required => true
method_option :reads, :type => :numeric, :required => true
method_option :samples, :type => :numeric, :required => true
def uc_filter
input = options[:input]
output = options[:output]
reads = options[:reads].to_i
samples = options[:samples].to_i
# load UC file
clstr_info = Helpers.load_uc_file input
clstr_counts = clstr_info[:clstr_counts] # clstr_counts[:clstr][sample.to_i] = reads
# filter
survivors = clstr_counts.reject do |a, b|
b.reject{ |i, j| j < reads }.length < samples
end
surviving_clusters = survivors.keys
# print filtered uc file
out = File.open(output, 'w')
kept, total = 0, 0
File.open(input) do |handle|
handle.each do |line|
if line =~ /^#/
out.print line
next
end
total += 1
if surviving_clusters.include? line.split[1].to_i
out.print line
kept += 1
end
end
end
out.close
ohai "Survivors"
ohai "clusters: #{surviving_clusters.length}/#{clstr_counts.keys.length} = #{100*surviving_clusters.length/clstr_counts.keys.length.to_f}%"
ohai "reads: #{kept}/#{total} = #{100*kept/total.to_f}%"
end
end
end
Version data entries
11 entries across 11 versions & 1 rubygems