bio-gadget 0.3.0 → 0.3.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 9bd8ea9d5f88557147da247ea9f817d4f069aae1
-  data.tar.gz: 60235e4a8b6b82c01366d27f38bd4c40e0ef0e71
+  metadata.gz: aaeb43dbf9adba74b5741250dab6e1764421a4c4
+  data.tar.gz: f45ff460eeee4aee51b7888f39592dac24075898
 SHA512:
-  metadata.gz: bdb115e0f1e9df39a39e108e7f8d00e9a1f2d69ed51b098217659050291034e5a052eaa2aed976cf6b9d18bea5a5929149367fe6b0f33e9a0ee7bc245668d2a5
-  data.tar.gz: d959b569cec30289b3bdf16676b4b26a5e0ed7adceb3e90407646ba074b5621d506eae98f362d57e4b838525d370fc91112e88141dea24a6ae4da3b2dc8faa80
+  metadata.gz: 0f7dbebd237bb629a0e1d758b6e024427586d54a7cf4ccac59270b34e9c2dec374ee0d5bf81c88465eaabf531a520c712876349d8e72db3793fd9560c45cceee
+  data.tar.gz: 4d39fb3ee8a65513bbea18750e435b5f5d54166c1e3f92920f6905e83248c7e0a39361382660ca80253572b466b0458d60b6cf4e3891971f02e34ef83292bed8

data/README.org

@@ -15,13 +15,13 @@ To check all commands in this package,
 
 Currently available commands are
 
-: bio
-: ---
-: bio-gadget dedup                                # deduplicate fastq (via STDIN)
-: bio-gadget demlt -b,--barcode-file=BARCODE-FILE # demultiplex fastq by barcodes
-: bio-gadget fqxz                                 # automatic (re)compression of *.fq(.gz|.bz2) files
-: bio-gadget qvstat QUAL                          # statistics of quality values in *.qual file
-: bio-gadget wigchr WIG CHR                       # extract wiggle track on specified chromosome
+: bio-gadget dedup                                 # deduplicate fastq (via STDIN)
+: bio-gadget demlt BARCODE [FASTQ]                 # demultiplex fastq by barcodes
+: bio-gadget ensann GENENAMEGZ [ENSGENE]           # create ensGene.gtf
+: bio-gadget fqxz                                  # automatic (re)compression of *.fq(.gz|.bz2) files
+: bio-gadget qvstat QUAL                           # statistics of quality values in *.qual file
+: bio-gadget ucscann ISOFORMSGZ XREFGZ [KNOWNGENE] # create knownGene.gtf
+: bio-gadget wigchr WIG CHR                        # extract wiggle track on specified chromosome
 
 * Contributing
 

data/lib/bio-gadget/demlt.rb

@@ -8,14 +8,12 @@ module Bio
 
     namespace :bio
 
-    desc 'demlt', 'demultiplex fastq from STDIN by barcodes'
-    option 'barcode-file', :aliases => '-b', :type => :string, :required => true
+    desc 'demlt BARCODE [FASTQ]', 'demultiplex fastq from STDIN by barcodes'
     option 'output-dir', :aliases => '-o', :type => :string, :default => '.'
     option 'umi-length', :aliases => '-u', :type => :numeric, :default => 4
     option 'cdna-length', :aliases => '-c', :type => :numeric, :default => 37
-    def demlt
+    def demlt(bcfile, fastq=:stdin)
 
-      bcfile = options['barcode-file']
       ofs = options['umi-length']
       len = options['cdna-length']
 
@@ -47,7 +45,7 @@ module Bio
         fifo1s = Array.new
         fifo1paths.each { |fifo1path| fifo1s.push(open(fifo1path, 'w')) }
         total = 0
-        Bio::Faster.new(:stdin).each_record(:quality => :raw) do |vals|
+        Bio::Faster.new(fastq).each_record(:quality => :raw) do |vals|
           fifo1 = fifo1s[total % procs]
           fifo1.puts(vals.join("\t"))
           total += 1

data/lib/bio-gadget/gtfann.rb

@@ -0,0 +1,93 @@
+module Bio
+  class Gadget < Thor
+
+    desc 'ucscann ISOFORMSGZ XREFGZ [KNOWNGENE]', 'create knownGene.gtf'
+    def ucscann(isoforms, xref, knowngene='/dev/stdin')
+
+      acc2id = Hash.new
+      open("| gunzip -c #{isoforms}").each { |line|
+        id, acc = line.rstrip.split(/\t/)
+        acc2id[acc] = id
+      }
+
+      acc2sym = Hash.new
+      open("| gunzip -c #{xref} | cut -f 1,5").each { |line|
+        acc, sym = line.rstrip.split(/\t/)
+        acc2sym[acc] = sym
+      }
+
+      open("| sort -k1,1 -k4,4n", 'w') { |fp|
+        open(knowngene).each { |line|
+          cols = line.rstrip.split(/\t/)
+          acc, chr, str = cols.values_at(0, 1, 2)
+          cs = cols[5].to_i
+          ce = cols[6].to_i
+          lefts = cols[8].split(/,/)
+          rights = cols[9].split(/,/)
+          prop = "gene_id \"#{acc2id[acc]}\"; transcript_id \"#{acc}\""
+          prop += "; gene_name \"#{acc2sym[acc]}\"" if acc2sym.has_key?(acc)
+          lefts.each_index { |i|
+            next if lefts[i].nil?
+            l = lefts[i].to_i
+            r = rights[i].to_i
+            fp.puts [chr, 'knownGene', 'exon', l+1, r, 0, str, '.', prop].join("\t")
+          }
+          if cs != ce
+            fp.puts [chr, 'knownGene', 'start_codon', (str == '+' ? cs+1 : ce-2), (str == '+' ? cs+3 : ce), 0, str, '.', prop].join("\t")
+            lefts.each_index { |i|
+              next if lefts[i].nil?
+              l = lefts[i].to_i
+              next if ce-1 < l
+              r = rights[i].to_i
+              next if r-1 < cs
+              fp.puts [chr, 'knownGene', 'CDS', (cs < l ? l : cs), (r < ce ? r : ce), 0, str, '.', prop].join("\t")
+            }
+          end
+        }
+      }
+
+    end
+
+    desc 'ensann GENENAMEGZ [ENSGENE]', 'create ensGene.gtf'
+    def ensann(genename, ensgene='/dev/stdin')
+
+      acc2sym = Hash.new
+      open("| gunzip -c #{genename}").each { |line|
+        acc, sym = line.rstrip.split(/\t/)
+        acc2sym[acc] = sym
+      }
+
+      open("| sort -k1,1 -k4,4n", 'w') { |fp|
+        open(ensgene).each { |line|
+          cols = line.rstrip.split(/\t/)
+          acc, chr, str, id = cols.values_at(1, 2, 3, 12)
+          cs = cols[6].to_i
+          ce = cols[7].to_i
+          lefts = cols[9].split(/,/)
+          rights = cols[10].split(/,/)
+          prop = "gene_id \"#{id}\"; transcript_id \"#{acc}\""
+          prop += "; gene_name \"#{acc2sym[acc]}\"" if acc2sym.has_key?(acc)
+          lefts.each_index { |i|
+            next if lefts[i].nil?
+            l = lefts[i].to_i
+            r = rights[i].to_i
+            fp.puts [chr, 'ensGene', 'exon', l+1, r, 0, str, '.', prop].join("\t")
+          }
+          if cs != ce
+            fp.puts [chr, 'ensGene', 'start_codon', (str == '+' ? cs+1 : ce-2), (str == '+' ? cs+3 : ce), 0, str, '.', prop].join("\t")
+            lefts.each_index { |i|
+              next if lefts[i].nil?
+              l = lefts[i].to_i
+              next if ce-1 < l
+              r = rights[i].to_i
+              next if r-1 < cs
+              fp.puts [chr, 'ensGene', 'CDS', (cs < l ? l : cs), (r < ce ? r : ce), 0, str, '.', prop].join("\t")
+            }
+          end
+        }
+      }
+
+    end
+
+  end
+end

data/lib/bio-gadget/version.rb

@@ -3,7 +3,7 @@ require 'thor'
 module Bio
   class Gadget < Thor
 
-    VERSION = "0.3.0"
+    VERSION = "0.3.1"
 
   end
 end

data/lib/bio-gadget.rb

@@ -2,6 +2,7 @@ require 'bio-gadget/version'
 require 'bio-gadget/dedup'
 require 'bio-gadget/demlt'
 require 'bio-gadget/fqxz'
+require 'bio-gadget/gtfann'
 require 'bio-gadget/qvstat'
 require 'bio-gadget/wigchr'
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: bio-gadget
 version: !ruby/object:Gem::Version
-  version: 0.3.0
+  version: 0.3.1
 platform: ruby
 authors:
 - Shintaro Katayama
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-03-16 00:00:00.000000000 Z
+date: 2013-03-17 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: thor
@@ -100,6 +100,7 @@ files:
 - lib/bio-gadget/dedup.rb
 - lib/bio-gadget/demlt.rb
 - lib/bio-gadget/fqxz.rb
+- lib/bio-gadget/gtfann.rb
 - lib/bio-gadget/qvstat.rb
 - lib/bio-gadget/version.rb
 - lib/bio-gadget/wigchr.rb