bio-gff3 0.8.4 → 0.8.5

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 8
-  - 4
-  version: 0.8.4
+  - 5
+  version: 0.8.5
 platform: ruby
 authors: 
 - Pjotr Prins
@@ -14,7 +14,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-01-03 00:00:00 +01:00
+date: 2011-01-13 00:00:00 +01:00
 default_executable: gff3-fetch
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -33,8 +33,23 @@ dependencies:
   prerelease: false
   version_requirements: *id001
 - !ruby/object:Gem::Dependency 
-  name: shoulda
+  name: bio-logger
   requirement: &id002 !ruby/object:Gem::Requirement 
+    none: false
+    requirements: 
+    - - ">"
+      - !ruby/object:Gem::Version 
+        segments: 
+        - 0
+        - 5
+        - 0
+        version: 0.5.0
+  type: :runtime
+  prerelease: false
+  version_requirements: *id002
+- !ruby/object:Gem::Dependency 
+  name: shoulda
+  requirement: &id003 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -44,10 +59,10 @@ dependencies:
         version: "0"
   type: :development
   prerelease: false
-  version_requirements: *id002
+  version_requirements: *id003
 - !ruby/object:Gem::Dependency 
   name: bundler
-  requirement: &id003 !ruby/object:Gem::Requirement 
+  requirement: &id004 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ~>
@@ -59,10 +74,10 @@ dependencies:
         version: 1.0.0
   type: :development
   prerelease: false
-  version_requirements: *id003
+  version_requirements: *id004
 - !ruby/object:Gem::Dependency 
   name: jeweler
-  requirement: &id004 !ruby/object:Gem::Requirement 
+  requirement: &id005 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ~>
@@ -74,10 +89,10 @@ dependencies:
         version: 1.5.2
   type: :development
   prerelease: false
-  version_requirements: *id004
+  version_requirements: *id005
 - !ruby/object:Gem::Dependency 
   name: rcov
-  requirement: &id005 !ruby/object:Gem::Requirement 
+  requirement: &id006 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
@@ -87,37 +102,81 @@ dependencies:
         version: "0"
   type: :development
   prerelease: false
-  version_requirements: *id005
+  version_requirements: *id006
+- !ruby/object:Gem::Dependency 
+  name: rspec
+  requirement: &id007 !ruby/object:Gem::Requirement 
+    none: false
+    requirements: 
+    - - ">="
+      - !ruby/object:Gem::Version 
+        segments: 
+        - 2
+        - 0
+        - 0
+        version: 2.0.0
+  type: :development
+  prerelease: false
+  version_requirements: *id007
 - !ruby/object:Gem::Dependency 
   name: bio
-  requirement: &id006 !ruby/object:Gem::Requirement 
+  requirement: &id008 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
         segments: 
         - 1
-        - 3
+        - 4
         - 1
-        version: 1.3.1
-  type: :development
+        version: 1.4.1
+  type: :runtime
   prerelease: false
-  version_requirements: *id006
+  version_requirements: *id008
 - !ruby/object:Gem::Dependency 
-  name: rspec
-  requirement: &id007 !ruby/object:Gem::Requirement 
+  name: log4r
+  requirement: &id009 !ruby/object:Gem::Requirement 
+    none: false
+    requirements: 
+    - - ">"
+      - !ruby/object:Gem::Version 
+        segments: 
+        - 1
+        - 1
+        - 6
+        version: 1.1.6
+  type: :runtime
+  prerelease: false
+  version_requirements: *id009
+- !ruby/object:Gem::Dependency 
+  name: bio-logger
+  requirement: &id010 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
         segments: 
-        - 2
         - 0
+        - 6
+        - 1
+        version: 0.6.1
+  type: :runtime
+  prerelease: false
+  version_requirements: *id010
+- !ruby/object:Gem::Dependency 
+  name: rspec
+  requirement: &id011 !ruby/object:Gem::Requirement 
+    none: false
+    requirements: 
+    - - ">"
+      - !ruby/object:Gem::Version 
+        segments: 
+        - 2
         - 0
-        version: 2.0.0
+        version: "2.0"
   type: :development
   prerelease: false
-  version_requirements: *id007
+  version_requirements: *id011
 description: |
   GFF3 (genome browser) information and digest mRNA and CDS sequences.
   Options for low memory use and caching of records.
@@ -141,20 +200,29 @@ files:
 - bin/gff3-fetch
 - bio-gff3.gemspec
 - lib/bio-gff3.rb
-- lib/bio/db/gff/gffassemble.rb
-- lib/bio/db/gff/gffdb.rb
-- lib/bio/db/gff/gfffasta.rb
-- lib/bio/db/gff/gfffileiterator.rb
-- lib/bio/db/gff/gffinmemory.rb
-- lib/bio/db/gff/gffnocache.rb
-- lib/bio/db/gff/gffparser.rb
+- lib/bio/db/gff/digest/gffinmemory.rb
+- lib/bio/db/gff/digest/gffnocache.rb
+- lib/bio/db/gff/digest/gffparser.rb
+- lib/bio/db/gff/file/gfffasta.rb
+- lib/bio/db/gff/file/gfffileiterator.rb
+- lib/bio/db/gff/gff3.rb
+- lib/bio/db/gff/gff3fasta.rb
+- lib/bio/db/gff/gff3parsefile.rb
+- lib/bio/db/gff/gff3parserec.rb
+- lib/bio/db/gff/gffcomponent.rb
+- lib/bio/db/gff/gffrecord.rb
+- lib/bio/db/gff/gffsection.rb
+- lib/bio/db/gff/gffsequence.rb
+- lib/bio/db/gff/gffvalidate.rb
 - lib/bio/output/gfffastawriter.rb
+- lib/bio/output/gfflogger.rb
 - lib/bio/system/lruhash.rb
 - spec/gff3_assemble2_spec.rb
 - spec/gff3_assemble3_spec.rb
 - spec/gff3_assemble_spec.rb
 - spec/gff3_fileiterator_spec.rb
 - spec/gffdb_spec.rb
+- spec/gffparserec.rb
 - test/data/gff/MhA1_Contig1133.fa
 - test/data/gff/MhA1_Contig1133.gff3
 - test/data/gff/MhA1_Contig125.fa
@@ -164,7 +232,12 @@ files:
 - test/data/gff/test-ext-fasta.fa
 - test/data/gff/test-ext-fasta.gff3
 - test/data/gff/test.gff3
+- test/data/regression/test_ext_gff3.rtest
+- test/data/regression/test_gff3.rtest
+- test/data/regression/test_nocache_ext_gff3.rtest
+- test/data/regression/test_nocache_gff3.rtest
 - test/helper.rb
+- test/regressiontest.rb
 - test/test_bio-gff3.rb
 has_rdoc: true
 homepage: http://github.com/pjotrp/bioruby-gff3
@@ -180,7 +253,6 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements: 
   - - ">="
     - !ruby/object:Gem::Version 
-      hash: 520620943
       segments: 
       - 0
       version: "0"
@@ -205,5 +277,7 @@ test_files:
 - spec/gff3_assemble_spec.rb
 - spec/gff3_fileiterator_spec.rb
 - spec/gffdb_spec.rb
+- spec/gffparserec.rb
 - test/helper.rb
+- test/regressiontest.rb
 - test/test_bio-gff3.rb

data/lib/bio/db/gff/gffassemble.rb

@@ -1,341 +0,0 @@
-#
-# = bio/db/gff/gffassemble.rb - Assemble mRNA and CDS from GFF
-#
-# Copyright::  Copyright (C) 2010
-#              Pjotr Prins <pjotr.prins@thebird.nl>
-# License::    The Ruby License
-#
-# Fetch information from a GFF file
-
-module Bio
-  module GFFbrowser
-
-    module Helpers
-
-      module Error
-        def info str, id=''
-          $stderr.print "Info: "+str+" <#{id}>\n"
-        end
-
-        def warn str, id=''
-          Kernel.warn "Warning: "+str+" <#{id}>"
-        end
-
-        def error str, id=''
-          Kernel.warn "Error: "+str+" <#{id}>"
-          exit(1) if $stop_on_error
-        end
-      end
-
-      # Helper class for counting IDs
-      class Counter < Hash
-        def add id
-          self[id] = 0 if self[id] == nil
-          self[id] += 1
-        end
-      end
-
-      # Helper class for storing linked records based on a shared ID
-      class LinkedRecs < Hash
-        include Error
-        def add id, rec
-          info "Adding #{rec.feature_type} <#{id}>"
-          self[id] = [] if self[id] == nil
-          self[id] << rec
-        end
-
-        # Validate all lists belong to the same container/component
-        def validate_seqname
-          each do | id, rec |
-            seqname = rec.first.seqname
-            rec.each do | section |
-              raise "Non-matching seqname #{section.seqname} in #{seqname}" if section.seqname != seqname
-            end
-          end
-        end
-     
-        # Validate all lists share the same parent (if available). First checks
-        # for Parent attribute, next for mRNA attribute
-        def validate_shared_parent
-          each do | id, rec |
-            parent = rec.first.get_attribute('Parent')
-            if parent
-              rec.each do | section |
-                _parent = section.get_attribute('Parent')
-                raise "Non-matching parent #{_parent} and #{parent} in #{id}" if _parent != parent
-              end
-            end
-            parent = rec.first.get_attribute('mRNA')
-            if parent
-              rec.each do | section |
-                _parent = section.get_attribute('mRNA')
-                raise "Non-matching parent #{_parent} and #{parent} in #{id}" if _parent != parent
-              end
-            end
-          end
-        end
-     
-        # walk all (CDS) lists for every container/component and 
-        # validate they do not overlap
-        def validate_nonoverlapping
-          each do | id, rec |
-            sections = Sections::sort(rec)
-            sections.each_with_index do | check, i |
-              neighbour = sections[i+1]
-              if neighbour and check.intersection(neighbour)
-                warn "Overlapping sections for ",id
-              end
-            end
-          end
-        end
-      end
-
-      class Section < Range
-        attr_reader :rec
-        def initialize rec
-          super(rec.start,rec.end)
-          @rec = rec
-        end
-        def intersection(other)
-          raise ArgumentError, 'value must be a Range' unless other.kind_of?(Range)
-          min, max = first, exclude_end? ? max : last
-          other_min, other_max = other.first, other.exclude_end? ? other.max : other.last
-          new_min = self === other_min ? other_min : other === min ? min : nil
-          new_max = self === other_max ? other_max : other === max ? max : nil
-          new_min && new_max ? new_min..new_max : nil
-        end
-        alias_method :&, :intersection
-        def <=> other
-          first <=> other.first
-        end
-      end
-
-      module Sections
-        # Return list of sorted Sections
-        def Sections::sort rec
-          sections = []
-          rec.each do | section |
-            sections.push Section.new(section)
-          end
-          sections.sort
-        end
-      end
-
-      module Record
-        include Error
-        # Format a record ID by, first, getting the ID attribute. If that fails
-        # the seqname is used with the start/stop positions.
-        def Record::formatID rec  
-          id = rec.id if rec.id
-          if !id
-            if rec.seqname
-              id = "#{rec.seqname} #{rec.start} #{rec.end}".strip
-            else
-              id = 'unknown'
-              $stderr.print "Record with unknown ID"+rec.to_s
-            end
-          end
-          id
-        end
-      end
-
-      module Gff3Component
-
-        include Error
-
-        COMPONENT_TYPES = %w{
-          gene SO:0000704 contig transcript Component region
-        }
- 
-        # Walk the component list to find a matching component/container for a
-        # record. First use the parent ID. If that is missing go by sequence
-        # name.
-        def find_component rec
-          parent = rec.get_attribute('Parent')
-          if @componentlist[parent] 
-            # nice, there is a match
-            info "find_component: Matched parent", parent
-            return @componentlist[parent]
-          end
-          search = rec.seqname
-          if @componentlist[search]
-            info "find_component: Matched seqname", search
-            return @componentlist[search] 
-          end
-          @componentlist.each do | componentid, component |
-            # dissemble id
-            (id, start, stop) = componentid.split(/ /)
-            if id==search and rec.start >= start.to_i and rec.end <= stop.to_i
-              info "find_component: Matched column 0 and location", componentid
-              return component
-            end
-          end
-          # Ah, painful. At this point the record has no matching container, probably
-          # because it has no parent ID and the component has an ID. We have to go by
-          # ID for every component individually
-          @componentlist.each do | componentid, component |
-            if component.seqname==search and rec.start >= component.start and rec.end <= component.end
-              # p ["----",search,rec]
-              # p component
-              info "find_component: Matched (long search) column 0 and location", componentid
-              return component
-            end
-          end
-          warn "Could not find container/component for",Record::formatID(rec)
-        end
-      end
-
-      module Gff3Features
-
-        # Ignore the following features (case sensitive?)
-        IGNORE_FEATURES = Gff3Component::COMPONENT_TYPES + %w{
-          transposon Match similarity UTR
-          TF_binding_site intronSO:0000188 polyA_sequence SO:0000610
-          polyA_site SO:0000553
-          five_prime_UTR SO:0000204 three_prime_UTR SO:0000205
-          exon SO:0000147
-        }
-      end
-
-      module Gff3Sequence
-        # Patch a sequence together from a Sequence string and an array
-        # of records. Note that rec positions are 1-based coordinates, relative 
-        # to the landmark given in column 1 - in this case the sequence as it
-        # is passed in. The following options are available:
-        #
-        #   :reverse      : do reverse if reverse is indicated (default true)
-        #   :complement   : do complement if reverse is indicated (default true)
-        #   :phase        : do set CDS phase (default false, normally ignore)
-        #   :trim         : make sure sequence is multiple of 3 nucleotide bps (default true)
-        #
-        # special options:
-        #
-        #   :raw          : raw sequence (all above false)
-        #   :codonize     : codon sequence (reverse, complement, and trim are true)
-        #   :fix          : fix errors (default false)
-        #
-        def assemble sequence, startpos, reclist, options = { :phase=>false, :reverse=>true, :trim=>true, :complement=>true, :fix=>false, :debug=>false }
-          # default to nil, if not passed in
-          do_debug = options[:debug]
-          do_phase = options[:phase]
-          do_fix        = options[:fix]
-          # default to true, if not passed in
-          do_reverse = (options[:reverse] == false ? false : true)
-          do_trim    = (options[:trim] == false ? false : true)
-          do_complement = (options[:complement] == false ? false : true)
-          if options[:raw]
-            do_phase = false
-            do_reverse = false
-            do_trim = false
-            do_complement = false
-          elsif options[:codonize]
-            do_phase = false
-            do_reverse = true
-            do_trim = true
-            do_complement = true
-          end
-          sectionlist = Sections::sort(reclist)
-          rec0 = sectionlist.first.rec
-          # we assume ORF is always read in the same direction
-          orf_reverse = (rec0.strand == '-')
-          orf_frame = startpos - 1
-          orf_frameshift = orf_frame % 3
-          sectionlist = sectionlist.reverse if orf_reverse
-          if do_debug
-            p "------------------"
-            p options 
-            p [:reverse,do_reverse]
-            p [:complement,do_complement]
-            p [:trim,do_trim]
-            p [:orf_reverse, orf_reverse, rec0.strand]
-          end
-
-          if sequence.kind_of?(Bio::FastaFormat)
-            # BioRuby conversion
-            sequence = sequence.seq
-          end
-          # Generate array of sequences
-          seq = sectionlist.map { | section |
-            rec = section.rec
-            s = sequence[(section.begin-1)..(section.end-1)]
-            if do_reverse and orf_reverse
-              s = s.reverse 
-            end
-            # Correct for phase. Unfortunately the use of phase is ambiguous.
-            # Here we check whether rec.start is in line with orf_frame. If it
-            # is, we correct for phase. Otherwise it is ignored.
-            if do_phase and rec.phase
-              phase = rec.phase.to_i
-              # if ((rec.start-startpos) % 3 == 0) 
-              s = s[phase..-1]
-              # end
-            end
-            s
-          }
-          # p seq
-          seq = seq.join
-          if do_complement and do_reverse and orf_reverse
-            ntseq = Bio::Sequence::NA.new(seq)
-            seq = ntseq.forward_complement.upcase
-          end
-          # This is the place to fix sequences (e.g. the Wormbase bug)
-          if do_fix or @options[:fix] or @options[:fix_wormbase]
-            if @options[:fix_wormbase] and rec0.id.index('gene1')==0
-              # Wormbase gene1 only, so ignore rest
-            else
-              test_frame = 0
-              ntseq = Bio::Sequence::NA.new(seq)
-              aaseq = ntseq.translate
-              if aaseq.count('*') > 1
-                test_frame = 1
-                seq = seq[1..-1] 
-                ntseq = Bio::Sequence::NA.new(seq)
-                aaseq = ntseq.translate
-                if aaseq.count('*') > 1
-                  test_frame = 2
-                  seq = seq[1..-1]
-                  ntseq = Bio::Sequence::NA.new(seq)
-                  aaseq = ntseq.translate
-                  raise 'Validation problem '+rec0.id if aaseq.count('*') > 1
-                end
-              end
-              if test_frame > 0
-                warn rec0.id,"Frame adjusted to #{test_frame} (fix)"
-              end
-            end
-          end
-          if do_trim
-            reduce = seq.size % 3
-            seq = seq[0..(seq.size-1-reduce)] if reduce != 0
-          end
-          if @options[:validate]
-            ntseq = Bio::Sequence::NA.new(seq)
-            aaseq = ntseq.translate
-            raise 'Validate translation problem '+rec0.id+"\n"+seq if aaseq.count('*') > 1
-          end
-
-          retval = seq
-          retval
-        end
-
-        # Patch a sequence together from a Sequence string and an array
-        # of records and translate in the correct direction and frame. The options 
-        # are the same as for +assemble+, except :trim defaults to true.
-        def assembleAA sequence, startpos, reclist, options = { :phase=>false, :reverse=>true, :trim=>true, :complement=>true }
-          seq = assemble(sequence, startpos, reclist, options)
-          ntseq = Bio::Sequence::NA.new(seq)
-          ntseq.translate
-        end
-
-        # Create a description for output
-        def description id, component, rec
-          sections = Sections::sort(rec)
-          id+' Sequence:'+component.seqname+"_#{component.start}:#{component.end} ("+
-           sections.map { |s| "#{s.first}:#{s.last}" }.join(', ')  +")"
-        end
-
-      end
-    end # Helpers
-
-  end
-end