lederhosen 0.3.4 → 0.3.5

data/lederhosen.gemspec

@@ -5,7 +5,7 @@
 
 Gem::Specification.new do |s|
   s.name = "lederhosen"
-  s.version = "0.3.4"
+  s.version = "0.3.5"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Austin G. Davis-Richardson"]

data/lib/lederhosen/tasks/otu_filter.rb

@@ -12,21 +12,26 @@ module Lederhosen
       input   = options[:input]
       output  = options[:output]
       reads   = options[:reads]
-      samples = options[:samples]
+      min_samples = options[:samples]
 
-      ohai "filtering otu file #{input} (reads = #{reads}, samples = #{samples}), saving to #{output}"
+      ohai "filtering otu file #{input} (reads = #{reads}, samples = #{min_samples}), saving to #{output}"
 
       ##
       # Iterate over otu table line by line.
       # Only print if cluster meets criteria
       #
-      kept = 0
+      kept_clusters = 0
+      total_reads   = 0
+      kept_reads    = 0
+
+      out = File.open(output, 'w')
+
       File.open(input) do |handle|
         header  = handle.gets.strip
         header  = header.split(',')
         samples = header[1..-1]
 
-        puts header.join(',')
+        out.puts header.join(',')
 
         handle.each do |line|
           line       = line.strip.split(',')
@@ -34,13 +39,17 @@ module Lederhosen
           counts     = line[1..-1].collect { |x| x.to_i }
 
           # should be the same as uc_filter
-          if counts.reject { |x| x < reads }.length > samples
-            puts line.join(',')
-            kept += 1
+          if counts.reject { |x| x < reads }.length > min_samples
+            out.puts line.join(',')
+            kept_clusters += 1
+            kept_reads += counts.inject(:+)
           end
+          total_reads += counts.inject(:+)
         end
       end
-      ohai "kept #{kept} clusters."
+
+      ohai "kept #{kept_reads} reads (#{kept_reads/total_reads.to_f})."
+      ohai "kept #{kept_clusters} clusters."
     end
 
   end

data/lib/lederhosen/version.rb

@@ -2,7 +2,7 @@ module Lederhosen
   module Version
     MAJOR = 0
     MINOR = 3
-    PATCH = 4
+    PATCH = 5
 
     STRING = [MAJOR, MINOR, PATCH].join('.')
   end

data/spec/cli_spec.rb

@@ -37,10 +37,14 @@ describe Lederhosen::CLI do
   end
 
   it 'should build OTU abundance matrices' do
-    `./bin/lederhosen otu_table --clusters=#{$test_dir}/clusters.uc --output=#{$test_dir}/test_tables`
+    `./bin/lederhosen otu_table --clusters=#{$test_dir}/clusters.uc --output=#{$test_dir}/otu_table.csv`
     $?.success?.should be_true
   end
 
+  it 'should filter OTU abundance matrices' do
+    `./bin/lederhosen otu_filter --input=#{$test_dir}/otu_table.csv --output=#{$test_dir}/otu_table.filtered.csv --reads 1 --samples 1`
+  end
+
   it 'should split joined.fasta into reads for each cluster' do
     `./bin/lederhosen split --reads=#{$test_dir}/joined.fasta --clusters=#{$test_dir}/clusters.uc --out-dir=#{$test_dir}/split --min-clst-size=1`
   end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: lederhosen
 version: !ruby/object:Gem::Version 
-  hash: 27
+  hash: 25
   prerelease: 
   segments: 
   - 0
   - 3
-  - 4
-  version: 0.3.4
+  - 5
+  version: 0.3.5
 platform: ruby
 authors: 
 - Austin G. Davis-Richardson