bio-gadget 0.4.6 → 0.4.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: dca3c35773b7e985b12c3e938d30bec303e0c5cf
-  data.tar.gz: d275ccd3b2f266af95406754876da216c0132237
+  metadata.gz: 993d87f1638e12bc336223e98be4dbeeeb5a8404
+  data.tar.gz: 95cb2d5cba5cc230d960bc0c558b3f1b1b562753
 SHA512:
-  metadata.gz: f2892f0584f058a6a3d640486b4526f59aa3608094bf8f11b481223944b7749ce90e877ee01a7e2e93300cbe8e6158f12499d8ba75ebc5f125cbfd6eb8b1db6c
-  data.tar.gz: 727a99f599f994d75fd1b0b1e32834b566d29819e41473e2421c564bbbed309a6c73aa8685498f52f920fdb9795a92ab2c2eef3f8a61e8ce297a683eccabd529
+  metadata.gz: ec8586c5e171b700596500bfbf16a9467bc65c6dd2dff2c5bd0e4b34f4775e40322cc1eecb00c81745c514edf3e59fedb8f9cb57a12586dcce4d08e6b87d24dd
+  data.tar.gz: da97c93be8331cf0048dd6e99cf6d66debc9985797ec909f6c2c908a6904667df2507e2170460632a5bd58a18dde44acd465843d03e572b4c55c9773fa9223f6

data/README.org

@@ -18,6 +18,7 @@ Currently available commands are
 - demlt :: Demultiplex fastq by barcodes
 - femrg :: Extract and merge first exons
 - fqxz :: (Re)compression of *.fq(.gz|.bz2) files
+- pead :: Find peak within each exon from wigs by a mojority vote.
 - qvstat :: Statistics of quality values in *.qual file
 - rgt2mtx :: Convert cuffdiff read group tracking file into tab-separated matrix
 - wig5p :: Convert bam-format alignments into wig-format table

data/lib/bio-gadget.rb

@@ -3,6 +3,7 @@ require 'bio-gadget/dedup'
 require 'bio-gadget/demlt'
 require 'bio-gadget/femrg'
 require 'bio-gadget/fqxz'
+require 'bio-gadget/peak'
 require 'bio-gadget/qvstat'
 require 'bio-gadget/rgt2mtx'
 require 'bio-gadget/wig5p'

data/lib/bio-gadget/peak.rb

@@ -0,0 +1,69 @@
+module Bio
+  class Gadget < Thor
+
+    desc 'peak WIG1,WIG2,... [GTF]', <<DESC
+Find peak within each exon from (gzipped) variableStep wigs by a majority vote. It will read from a standard input if no GTF option.
+DESC
+    def peak(wigs, gtf="/dev/stdin")
+
+      nchrpos2val = Hash.new
+      wigs.split(/,/).each { |wig|
+        n = wigs.split(/,/).index(wig)
+        nchrpos2val[n] = Hash.new unless nchrpos2val.key?(n)
+        chr = ''
+        fp = open("| zcat #{wig} | grep -v ^track")
+        fp.each { |line|
+          cols = line.rstrip.split(/\s+/)
+          if cols[0] == 'variableStep'
+            chr = cols[1].match(/chrom=(\S+)$/).to_a[1]
+            nchrpos2val[n][chr] = Hash.new unless nchrpos2val[n].key?(chr)
+          else
+            nchrpos2val[n][chr][cols[0].to_i] = cols[1].to_f
+          end
+        }
+        fp.close
+      }
+
+      open("| grep exon #{gtf}").each { |line|
+        cols = line.rstrip.split(/\t/)
+        oid = cols[8].match(/oId \"([^\"]+)/).to_a[1]
+        oid = "#{cols[0]}.#{oid}" if cols[0] =~ /^RNA_SPIKE_/
+        exn = cols[8].match(/exon_number \"(\d+)\"/).to_a[1]
+        chr = cols[0]
+        str = cols[6]
+        start = cols[3].to_i
+        stop = cols[4].to_i
+        peak = ''
+        #
+        poss = Hash.new
+        nchrpos2val.each { |n, chrpos2val|
+          if chrpos2val.key?(chr)
+            pos2val = chrpos2val[chr]
+            tmppos2val = Hash.new
+            pos2val.each { |pos, val|
+              tmppos2val[pos] = val if start <= pos && pos <= stop
+            }
+            if tmppos2val.size > 0
+              tmpposs = tmppos2val.keys.sort { |a, b|
+                tmppos2val[b] == tmppos2val[a] ? (str == '+' ? (a <=> b) : (b <=> a)) : (tmppos2val[b] <=> tmppos2val[a])
+              }
+              tmppos = tmpposs[0]
+              # puts "#{n} | #{chr}:#{start}-#{stop} #{str} | #{tmpposs}"
+              poss[tmppos] = poss.key?(tmppos) ? poss[tmppos]+1 : 1
+            end
+          end
+        }
+        if poss.size > 0
+          peaks = poss.keys.sort { |a, b|
+            poss[b] == poss[a] ? (str == '+' ? (a <=> b) : (b <=> a)) : (poss[b] <=> poss[a])
+          }
+          peak = peaks[0]
+        end
+        #
+        puts [oid, exn, peak].join("\t")
+      }
+
+    end
+
+  end
+end

data/lib/bio-gadget/version.rb

@@ -3,7 +3,7 @@ require 'thor'
 module Bio
   class Gadget < Thor
 
-    VERSION = "0.4.6"
+    VERSION = "0.4.7"
 
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: bio-gadget
 version: !ruby/object:Gem::Version
-  version: 0.4.6
+  version: 0.4.7
 platform: ruby
 authors:
 - Shintaro Katayama
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-05-03 00:00:00.000000000 Z
+date: 2013-05-30 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: thor
@@ -101,6 +101,7 @@ files:
 - lib/bio-gadget/demlt.rb
 - lib/bio-gadget/femrg.rb
 - lib/bio-gadget/fqxz.rb
+- lib/bio-gadget/peak.rb
 - lib/bio-gadget/qvstat.rb
 - lib/bio-gadget/rgt2mtx.rb
 - lib/bio-gadget/version.rb