bix 0.0.8 → 0.0.9

data/lib/bix.rb

@@ -6,4 +6,5 @@ require "#{File.dirname(__FILE__)}/bix/fastq"
 require "#{File.dirname(__FILE__)}/bix/fasta"
 require "#{File.dirname(__FILE__)}/bix/blast"
 require "#{File.dirname(__FILE__)}/bix/gtf"
+require "#{File.dirname(__FILE__)}/bix/plink"
 

data/lib/bix/plink.rb

@@ -0,0 +1,89 @@
+require 'csv'
+
+module Bix
+  # Reads a tfam file, extracting a list of the individual names
+  def self.read_tfam(io_or_fn)
+    io = io_or_fn.instance_of?(String) ? File.new(io_or_fn) : io
+
+    inds = []
+
+    for line in io
+      line.chomp!
+      next if line[0] == '#'
+      f = line.split(/\t| /)
+      inds << f[1]
+    end
+
+    return inds
+  end
+
+  # Reads a tped file. Produces a hash of ind_name => [haplotype 1, haplotype 2].
+  # ind_list specifies ind_names, if nil, ind names are just 0,1,2...
+  # as_array indicates whether haps will be output as strings or arrays
+  def self.read_tped(io_or_fn, ind_list=nil, as_array=false)
+    io = io_or_fn.instance_of?(String) ? File.new(io_or_fn) : io
+
+    ind_map = nil
+
+    for line in io
+      line.chomp!
+      next if line[0] == '#'
+      f = line.split(/ |\t/)
+
+      if !ind_map
+        ind_map = {}
+        # Initialize ind_map first
+        if ind_list
+          for i in 0...ind_list.size
+            ind_map[ind_list[i]] = i
+          end
+        else
+          ((f.size-4)/2).times do |ind_idx|
+            ind_map[ind_idx] = ind_idx
+          end
+        end
+
+        # Initialize haps object
+        haps = {} # name => [hap1_vec, hap2_vec]
+        for name, ind_idx in ind_map
+          if as_array
+            haps[name] = [[], []]
+          else
+            haps[name] = ["", ""]
+          end
+        end
+      end
+
+
+      for name, ind_idx in ind_map
+        haps[name][0] << f[4 + 2 * ind_idx + 0]
+        haps[name][1] << f[4 + 2 * ind_idx + 1]
+      end
+    end
+
+    return haps
+  end
+
+  # Reads a frq file, producing a list of hashes with the keys
+  # :chr, :snp, :a1, :a2, :maf, :nchrobs
+  def self.read_frq(io_or_fn)
+    io = io_or_fn.instance_of?(String) ? File.new(io_or_fn) : io
+    io.gets # header
+    res = []
+    for line in io
+      f = line.chomp.split
+      h = {}
+      h[:chr] = f[0]
+      h[:snp] = f[1]
+      h[:a1] = f[2]
+      h[:a2] = f[3]
+      h[:maf] = f[4].to_f
+      h[:nchrobs] = f[5].to_i
+
+      res << h
+    end
+
+    return res
+  end
+
+end

metadata

@@ -2,7 +2,7 @@
 name: bix
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 0.0.8
+  version: 0.0.9
 platform: ruby
 authors: 
 - Jesse Rodriguez
@@ -10,7 +10,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2013-01-21 00:00:00 Z
+date: 2013-07-11 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: swak
@@ -57,6 +57,7 @@ files:
 - lib/bix.rb
 - lib/bix/fasta.rb
 - lib/bix/fastq.rb
+- lib/bix/plink.rb
 - lib/bix/blast.rb
 - lib/bix/gtf.rb
 homepage: https://github.com/jesserod/bixruby