scaffolder-annotation-locator 0.0.1 → 0.1.0

data/Gemfile

@@ -1,7 +1,7 @@
 source "http://rubygems.org"
 
 group :default do
-  gem "scaffolder", "~> 0.4"
+  gem "scaffolder", "0.4.3"
 end
 
 group :development do
@@ -9,7 +9,7 @@ group :development do
   gem "jeweler",  "~> 1.5"
 
   gem "rspec",                   "~> 2.4"
-  gem "scaffolder-test-helpers",    "0.2.2"
+  gem "scaffolder-test-helpers",    "0.3.0"
   gem "cucumber",                "~> 0.9"
   gem "aruba",                   "~> 0.2"
 

data/VERSION

@@ -1 +1 @@
-0.0.1
+0.1.0

data/features/inserts.feature

@@ -0,0 +1,128 @@
+Feature: Parsing contigs with inserts
+  In order to include inserts in a scaffold
+  A user can use scaffold-annotation-locator
+  to update annotation coordinates with respect to contigs with inserts
+
+  Scenario: An annotation before an insert in a contig
+    Given a file named "scaf.yml" with:
+      """
+      ---
+        - sequence:
+            source: contig1
+            inserts:
+            -
+              source: insert1
+              open: 14
+              close: 15
+      """
+    Given a file named "seq.fna" with:
+      """
+      > contig1
+      AAAAAGGGGGCCCCCTTTTT
+      > insert1
+      TTTT
+      """
+    Given a file named "anno.gff" with:
+      """
+      ##gff-version 3
+      contig1	.	CDS	4	13	.	+	1	ID=gene1
+      """
+    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
+    Then the result should be:
+      """
+      ##gff-version 3
+      scaffold	.	CDS	4	13	.	+	1	ID=gene1
+      """
+
+  Scenario: An annotation after an insert in a contig
+    Given a file named "scaf.yml" with:
+      """
+      ---
+        - sequence:
+            source: contig1
+            inserts:
+            -
+              source: insert1
+              open: 1
+              close: 3
+      """
+    Given a file named "seq.fna" with:
+      """
+      > contig1
+      AAAAAGGGGGCCCCCTTTTT
+      > insert1
+      TTTT
+      """
+    Given a file named "anno.gff" with:
+      """
+      ##gff-version 3
+      contig1	.	CDS	4	13	.	+	1	ID=gene1
+      """
+    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
+    Then the result should be:
+      """
+      ##gff-version 3
+      scaffold	.	CDS	5	14	.	+	1	ID=gene1
+      """
+
+  Scenario: An annotation before an insert in a reversed contig
+    Given a file named "scaf.yml" with:
+      """
+      ---
+        - sequence:
+            source: contig1
+            reverse: true
+            inserts:
+            -
+              source: insert1
+              open: 1
+              close: 3
+      """
+    Given a file named "seq.fna" with:
+      """
+      > contig1
+      AAAAAGGGGGCCCCCTTTTT
+      > insert1
+      TTTT
+      """
+    Given a file named "anno.gff" with:
+      """
+      ##gff-version 3
+      contig1	.	CDS	4	13	.	+	1	ID=gene1
+      """
+    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
+    Then the result should be:
+      """
+      ##gff-version 3
+      scaffold	.	CDS	8	17	.	-	1	ID=gene1
+      """
+
+  Scenario: An annotation overlapping with an insert location
+    Given a file named "scaf.yml" with:
+      """
+      ---
+        - sequence:
+            source: contig1
+            inserts:
+            -
+              source: insert1
+              open: 1
+              close: 4
+      """
+    Given a file named "seq.fna" with:
+      """
+      > contig1
+      AAAAAGGGGGCCCCCTTTTT
+      > insert1
+      TTTT
+      """
+    Given a file named "anno.gff" with:
+      """
+      ##gff-version 3
+      contig1	.	CDS	4	13	.	+	1	ID=gene1
+      """
+    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
+    Then the result should be:
+      """
+      ##gff-version 3
+      """

data/features/{gff3.feature → multiple-contigs.feature}

@@ -1,81 +1,7 @@
-Feature: Locating gff3 annotations on a scaffold
-  In order to add gff3 annotations to a scaffold
+Feature: Locating annotations on single contig scaffold
+  In order to build a genome from multiple contigs
   A user can use scaffold-annotation-locator
-  to return the updated coordinates of scaffold annotations
-
-  Scenario: One annotation on a contig
-    Given a file named "scaf.yml" with:
-      """
-      ---
-        - sequence:
-            source: contig1
-      """
-    Given a file named "seq.fna" with:
-      """
-      > contig1
-      AAAAAGGGGGCCCCCTTTTT
-      """
-    Given a file named "anno.gff" with:
-      """
-      ##gff-version 3
-      contig1	.	CDS	4	13	.	+	1	ID=gene1
-      """
-    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
-    Then the result should be:
-      """
-      ##gff-version 3
-      scaffold	.	CDS	4	13	.	+	1	ID=gene1
-      """
-
-  Scenario: One annotation on a trimmed contig
-    Given a file named "scaf.yml" with:
-      """
-      ---
-        - sequence:
-            source: contig1
-            start: 4
-      """
-    Given a file named "seq.fna" with:
-      """
-      > contig1
-      AAAAAGGGGGCCCCCTTTTT
-      """
-    Given a file named "anno.gff" with:
-      """
-      ##gff-version 3
-      contig1	.	CDS	4	13	.	+	1	ID=gene1
-      """
-    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
-    Then the result should be:
-      """
-      ##gff-version 3
-      scaffold	.	CDS	1	10	.	+	1	ID=gene1
-      """
-
-  Scenario: One annotation on a reversed contig
-    Given a file named "scaf.yml" with:
-      """
-      ---
-        - sequence:
-            source: contig1
-            reverse: true
-      """
-    Given a file named "seq.fna" with:
-      """
-      > contig1
-      AAAAAGGGGGCCCCCTTTTT
-      """
-    Given a file named "anno.gff" with:
-      """
-      ##gff-version 3
-      contig1	.	CDS	1	6	.	+	1	ID=gene1
-      """
-    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
-    Then the result should be:
-      """
-      ##gff-version 3
-      scaffold	.	CDS	15	20	.	-	1	ID=gene1
-      """
+  to update annotation coordinates with respect from multiple contigs
 
   Scenario: Three annotations on three contigs
     Given a file named "scaf.yml" with:
@@ -257,61 +183,80 @@ Feature: Locating gff3 annotations on a scaffold
       scaffold	.	CDS	41	46	.	+	1	ID=gene2
       """
 
-  Scenario: Annotations on two contigs separated by an unresolved region
+  Scenario: Annotations on a single duplicated contig
     Given a file named "scaf.yml" with:
       """
       ---
         - sequence:
             source: contig1
-        - unresolved:
-            length: 10
         - sequence:
-            source: contig2
+            source: contig1
       """
     Given a file named "seq.fna" with:
       """
       > contig1
       AAAAAGGGGGCCCCCTTTTT
-      > contig2
-      AAAAAGGGGGCCCCCTTTTT
       """
     Given a file named "anno.gff" with:
       """
       ##gff-version 3
       contig1	.	CDS	1	6	.	+	1	ID=gene1
-      contig2	.	CDS	1	6	.	+	1	ID=gene2
       """
     When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
     Then the result should be:
       """
       ##gff-version 3
       scaffold	.	CDS	1	6	.	+	1	ID=gene1
-      scaffold	.	CDS	31	36	.	+	1	ID=gene2
+      scaffold	.	CDS	21	26	.	+	1	ID=gene1
       """
 
-  Scenario: Annotations on a single duplicated contig
+  Scenario: Annotations on reversed and trimmed contigs with inserts
     Given a file named "scaf.yml" with:
       """
       ---
         - sequence:
             source: contig1
+            stop: 6
         - sequence:
-            source: contig1
+            source: contig2
+            reverse: true
+            inserts:
+            - 
+              source: insert1
+              open: 6
+              close: 7
+        - sequence:
+            source: contig3
+            start: 3
+
       """
     Given a file named "seq.fna" with:
       """
       > contig1
-      AAAAAGGGGGCCCCCTTTTT
+      AAAAAGGG
+      > contig2
+      AAAAAGGGGGC
+      > contig3
+      AAAAAGGG
+      > insert1
+      TTT
       """
     Given a file named "anno.gff" with:
       """
       ##gff-version 3
-      contig1	.	CDS	1	6	.	+	1	ID=gene1
+      contig1	.	CDS	1	4	.	+	1	ID=gene1
+      contig1	.	CDS	5	8	.	+	1	ID=gene2
+      contig2	.	CDS	1	4	.	+	1	ID=gene3
+      contig2	.	CDS	8	11	.	+	1	ID=gene4
+      contig3	.	CDS	1	3	.	+	1	ID=gene5
+      contig3	.	CDS	4	8	.	+	1	ID=gene6
       """
     When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
     Then the result should be:
       """
       ##gff-version 3
-      scaffold	.	CDS	1	6	.	+	1	ID=gene1
-      scaffold	.	CDS	21	26	.	+	1	ID=gene1
+      scaffold	.	CDS	1	4	.	+	1	ID=gene1
+      scaffold	.	CDS	15	18	.	-	1	ID=gene3
+      scaffold	.	CDS	7	10	.	-	1	ID=gene4
+      scaffold	.	CDS	20	24	.	+	1	ID=gene6
       """

data/features/single-contig.feature

@@ -0,0 +1,158 @@
+Feature: Locating annotations on single contig scaffold
+  In order to add gff3 annotations to a scaffold
+  A user can use scaffold-annotation-locator
+  to return the updated coordinates of scaffold annotations
+
+  Scenario: One annotation on a contig
+    Given a file named "scaf.yml" with:
+      """
+      ---
+        - sequence:
+            source: contig1
+      """
+    Given a file named "seq.fna" with:
+      """
+      > contig1
+      AAAAAGGGGGCCCCCTTTTT
+      """
+    Given a file named "anno.gff" with:
+      """
+      ##gff-version 3
+      contig1	.	CDS	4	13	.	+	1	ID=gene1
+      """
+    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
+    Then the result should be:
+      """
+      ##gff-version 3
+      scaffold	.	CDS	4	13	.	+	1	ID=gene1
+      """
+
+  Scenario: One annotation on a reversed contig
+    Given a file named "scaf.yml" with:
+      """
+      ---
+        - sequence:
+            source: contig1
+            reverse: true
+      """
+    Given a file named "seq.fna" with:
+      """
+      > contig1
+      AAAAAGGGGGCCCCCTTTTT
+      """
+    Given a file named "anno.gff" with:
+      """
+      ##gff-version 3
+      contig1	.	CDS	1	6	.	+	1	ID=gene1
+      """
+    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
+    Then the result should be:
+      """
+      ##gff-version 3
+      scaffold	.	CDS	15	20	.	-	1	ID=gene1
+      """
+
+  Scenario: An annotation in a start trimmed region of the sequence
+    Given a file named "scaf.yml" with:
+      """
+      ---
+        - sequence:
+            source: contig1
+            start: 5
+      """
+    Given a file named "seq.fna" with:
+      """
+      > contig1
+      AAAAAGGGGGCCCCCTTTTT
+      > insert1
+      TTTT
+      """
+    Given a file named "anno.gff" with:
+      """
+      ##gff-version 3
+      contig1	.	CDS	4	13	.	+	1	ID=gene1
+      """
+    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
+    Then the result should be:
+      """
+      ##gff-version 3
+      """
+
+  Scenario: An annotation inside a stop trimmed region of the sequence
+    Given a file named "scaf.yml" with:
+      """
+      ---
+        - sequence:
+            source: contig1
+            stop: 12
+      """
+    Given a file named "seq.fna" with:
+      """
+      > contig1
+      AAAAAGGGGGCCCCCTTTTT
+      > insert1
+      TTTT
+      """
+    Given a file named "anno.gff" with:
+      """
+      ##gff-version 3
+      contig1	.	CDS	4	13	.	+	1	ID=gene1
+      """
+    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
+    Then the result should be:
+      """
+      ##gff-version 3
+      """
+
+  Scenario: An annotation bordering a stop trimmed region of the sequence
+    Given a file named "scaf.yml" with:
+      """
+      ---
+        - sequence:
+            source: contig1
+            stop: 13
+      """
+    Given a file named "seq.fna" with:
+      """
+      > contig1
+      AAAAAGGGGGCCCCCTTTTT
+      > insert1
+      TTTT
+      """
+    Given a file named "anno.gff" with:
+      """
+      ##gff-version 3
+      contig1	.	CDS	4	13	.	+	1	ID=gene1
+      """
+    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
+    Then the result should be:
+      """
+      ##gff-version 3
+      scaffold	.	CDS	4	13	.	+	1	ID=gene1
+      """
+
+  Scenario: An annotation bordering a start trimmed region of the sequence
+    Given a file named "scaf.yml" with:
+      """
+      ---
+        - sequence:
+            source: contig1
+            start: 4
+      """
+    Given a file named "seq.fna" with:
+      """
+      > contig1
+      AAAAAGGGGGCCCCCTTTTT
+      """
+    Given a file named "anno.gff" with:
+      """
+      ##gff-version 3
+      contig1	.	CDS	4	13	.	+	1	ID=gene1
+      """
+    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
+    Then the result should be:
+      """
+      ##gff-version 3
+      scaffold	.	CDS	1	10	.	+	1	ID=gene1
+      """
+

data/features/unresolved.feature

@@ -0,0 +1,37 @@
+Feature: Parsing unresolved regions
+  In order to include unresolved regions in a scaffold
+  A user can use scaffold-annotation-locator
+  to update annotation coordinates with respect to unresolved regions
+
+  Scenario: Annotations on two contigs separated by an unresolved region
+    Given a file named "scaf.yml" with:
+      """
+      ---
+        - sequence:
+            source: contig1
+        - unresolved:
+            length: 10
+        - sequence:
+            source: contig2
+      """
+    Given a file named "seq.fna" with:
+      """
+      > contig1
+      AAAAAGGGGGCCCCCTTTTT
+      > contig2
+      AAAAAGGGGGCCCCCTTTTT
+      """
+    Given a file named "anno.gff" with:
+      """
+      ##gff-version 3
+      contig1	.	CDS	1	6	.	+	1	ID=gene1
+      contig2	.	CDS	1	6	.	+	1	ID=gene2
+      """
+    When I relocate the annotations using "scaf.yml", "seq.fna" and "anno.gff"
+    Then the result should be:
+      """
+      ##gff-version 3
+      scaffold	.	CDS	1	6	.	+	1	ID=gene1
+      scaffold	.	CDS	31	36	.	+	1	ID=gene2
+      """
+

data/lib/scaffolder/annotation_locator.rb

@@ -2,6 +2,8 @@ require 'delegate'
 require 'scaffolder'
 require 'bio'
 
+require 'scaffolder/extensions'
+
 class Scaffolder::AnnotationLocator < DelegateClass(Array)
 
   def initialize(scaffold_file,sequence_file,gff_file)
@@ -10,37 +12,42 @@ class Scaffolder::AnnotationLocator < DelegateClass(Array)
     @gff_file      = gff_file
 
     updated_records = Array.new
-    scaffold.inject(0) do |length,entry|
+    scaffold.inject(0) do |prior_length,entry|
 
       if entry.entry_type == :sequence
-        updated_records << records[entry.source].map do |record|
-          update_record(record,entry,length)
-        end
-      end
+        records[entry.source].each do |record|
 
-      length + entry.sequence.length
-    end
+          # Don't include this record if it overlaps with an insert
+          next if record.overlap?(entry.inserts.map{|i| (i.open..i.close)})
 
-    super updated_records.flatten
-  end
+          # Skip this record it lies in the start or stop trimmed regions
+          next if record.start < entry.start
+          next if record.end   > entry.stop
 
-  def update_record(record,scaffold_entry,prior_length)
-    record.start -= scaffold_entry.start - 1
-    record.end   -= scaffold_entry.start - 1
+          # Update record location by size differences of prior inserts
+          entry.inserts.select {|i| i.close < record.start }.each do |insert|
+            record.change_position_by insert.size_diff
+          end
 
-    if scaffold_entry.reverse
-      record.end   = scaffold_entry.sequence.length - (record.end - 1)
-      record.start = scaffold_entry.sequence.length - (record.start - 1)
+          # Decrease record position by distance contig is trimmed at start
+          record.change_position_by(1 - entry.start)
 
-      record.end, record.start = record.start, record.end
-      record.strand = self.class.flip_strand(record.strand)
-    end
+          # Reverse complement record positions if contig is reversed
+          record.reverse_complement_by entry.sequence.length if entry.reverse
+
+          # Increase record position by length of prior contigs
+          record.change_position_by prior_length
 
-    record.start += prior_length
-    record.end   += prior_length
+          record.seqname = "scaffold"
 
-    record.seqname = "scaffold"
-    record
+          updated_records << record
+        end
+      end
+
+      prior_length + entry.sequence.length
+    end
+
+    super updated_records
   end
 
   def scaffold
@@ -55,8 +62,4 @@ class Scaffolder::AnnotationLocator < DelegateClass(Array)
     end
   end
 
-  def self.flip_strand(strand)
-    strand == '+' ? '-' : '+'
-  end
-
 end

data/lib/scaffolder/extensions.rb

@@ -0,0 +1,3 @@
+require 'scaffolder/gff_record_helper'
+
+Bio::GFF::GFF3::Record.send(:include, Scaffolder::GffRecordHelper)

data/lib/scaffolder/gff_record_helper.rb

@@ -0,0 +1,26 @@
+module Scaffolder::GffRecordHelper
+
+  def flip_strand
+    self.strand = (self.strand == '+' ? '-' : '+')
+  end
+
+  def change_position_by(distance)
+    self.start += distance
+    self.end += distance
+  end
+
+  def reverse_complement_by(distance)
+    self.end   = distance - (self.end - 1)
+    self.start = distance - (self.start - 1)
+
+    self.end, self.start = self.start, self.end
+    self.flip_strand
+  end
+
+  def overlap?(*ranges)
+    ranges.flatten.any? do |range|
+      range.include?(self.start) || range.include?(self.end)
+    end
+  end
+
+end

data/scaffolder-annotation-locator.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{scaffolder-annotation-locator}
-  s.version = "0.0.1"
+  s.version = "0.1.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Michael Barton"]
-  s.date = %q{2011-04-05}
+  s.date = %q{2011-06-10}
   s.description = %q{Build a genome scaffold using scaffolder and a set of annotated contigs. This tool updates the locations of the contig annotations using the scaffolder tempalte as a base.}
   s.email = %q{mail@michaelbarton.me.uk}
   s.extra_rdoc_files = [
@@ -23,12 +23,18 @@ Gem::Specification.new do |s|
     "README.rdoc",
     "Rakefile",
     "VERSION",
-    "features/gff3.feature",
+    "features/inserts.feature",
+    "features/multiple-contigs.feature",
+    "features/single-contig.feature",
     "features/step_definitions/scaffolder-annotation-locator_steps.rb",
     "features/support/env.rb",
+    "features/unresolved.feature",
     "lib/scaffolder/annotation_locator.rb",
+    "lib/scaffolder/extensions.rb",
+    "lib/scaffolder/gff_record_helper.rb",
     "scaffolder-annotation-locator.gemspec",
     "spec/scaffolder/annotation_locator_spec.rb",
+    "spec/scaffolder/gff_record_helper_spec.rb",
     "spec/spec_helper.rb",
     "spec/support/gff_attribute_matcher.rb"
   ]
@@ -37,41 +43,36 @@ Gem::Specification.new do |s|
   s.require_paths = ["lib"]
   s.rubygems_version = %q{1.3.7}
   s.summary = %q{Update locations of gff3 annotations from a scaffolder template}
-  s.test_files = [
-    "spec/scaffolder/annotation_locator_spec.rb",
-    "spec/spec_helper.rb",
-    "spec/support/gff_attribute_matcher.rb"
-  ]
 
   if s.respond_to? :specification_version then
     current_version = Gem::Specification::CURRENT_SPECIFICATION_VERSION
     s.specification_version = 3
 
     if Gem::Version.new(Gem::VERSION) >= Gem::Version.new('1.2.0') then
-      s.add_runtime_dependency(%q<scaffolder>, ["~> 0.4"])
+      s.add_runtime_dependency(%q<scaffolder>, ["= 0.4.3"])
       s.add_development_dependency(%q<bundler>, ["~> 1.0"])
       s.add_development_dependency(%q<jeweler>, ["~> 1.5"])
       s.add_development_dependency(%q<rspec>, ["~> 2.4"])
-      s.add_development_dependency(%q<scaffolder-test-helpers>, ["= 0.2.2"])
+      s.add_development_dependency(%q<scaffolder-test-helpers>, ["= 0.3.0"])
       s.add_development_dependency(%q<cucumber>, ["~> 0.9"])
       s.add_development_dependency(%q<aruba>, ["~> 0.2"])
       s.add_development_dependency(%q<yard>, ["~> 0.6"])
     else
-      s.add_dependency(%q<scaffolder>, ["~> 0.4"])
+      s.add_dependency(%q<scaffolder>, ["= 0.4.3"])
       s.add_dependency(%q<bundler>, ["~> 1.0"])
       s.add_dependency(%q<jeweler>, ["~> 1.5"])
       s.add_dependency(%q<rspec>, ["~> 2.4"])
-      s.add_dependency(%q<scaffolder-test-helpers>, ["= 0.2.2"])
+      s.add_dependency(%q<scaffolder-test-helpers>, ["= 0.3.0"])
       s.add_dependency(%q<cucumber>, ["~> 0.9"])
       s.add_dependency(%q<aruba>, ["~> 0.2"])
       s.add_dependency(%q<yard>, ["~> 0.6"])
     end
   else
-    s.add_dependency(%q<scaffolder>, ["~> 0.4"])
+    s.add_dependency(%q<scaffolder>, ["= 0.4.3"])
     s.add_dependency(%q<bundler>, ["~> 1.0"])
     s.add_dependency(%q<jeweler>, ["~> 1.5"])
     s.add_dependency(%q<rspec>, ["~> 2.4"])
-    s.add_dependency(%q<scaffolder-test-helpers>, ["= 0.2.2"])
+    s.add_dependency(%q<scaffolder-test-helpers>, ["= 0.3.0"])
     s.add_dependency(%q<cucumber>, ["~> 0.9"])
     s.add_dependency(%q<aruba>, ["~> 0.2"])
     s.add_dependency(%q<yard>, ["~> 0.6"])

data/spec/scaffolder/annotation_locator_spec.rb

@@ -5,7 +5,7 @@ describe Scaffolder::AnnotationLocator do
   def relocate(scaffold,records)
     @scaffold_file, @sequence_file = generate_scaffold_files(scaffold)
     described_class.new(@scaffold_file.path, @sequence_file.path,
-                        generate_gff3_file(records))
+                        generate_gff3_file(records).path)
   end
 
   before do
@@ -73,6 +73,115 @@ describe Scaffolder::AnnotationLocator do
 
     end
 
+    describe "with an insert before an annotation" do
+
+      subject do
+        relocate([@contig.clone.inserts(:open => 1, :close => 2, :sequence => 'TTT')],
+                 [@record])
+      end
+
+      it{ should set_the_attribute(:seqname => 'scaffold') }
+      it{ should set_the_attribute(:phase   => 1) }
+      it{ should set_the_attribute(:strand  => '+') }
+
+      it{ should set_the_attribute(:start   => 5).only_for_the(:first) }
+      it{ should set_the_attribute(:end     => 7).only_for_the(:first) }
+
+    end
+
+    describe "with an insert after an annotation" do
+
+      subject do
+        relocate([@contig.clone.
+                   inserts(:open => 7, :close => 8, :sequence => 'TTT').
+                   sequence('ATGTTTCCC')],
+                 [@record])
+      end
+
+      it{ should set_the_attribute(:seqname => 'scaffold') }
+      it{ should set_the_attribute(:phase   => 1) }
+      it{ should set_the_attribute(:strand  => '+') }
+
+      it{ should set_the_attribute(:start   => 4).only_for_the(:first) }
+      it{ should set_the_attribute(:end     => 6).only_for_the(:first) }
+
+    end
+
+    describe "with an insert before and after an annotation" do
+
+      subject do
+        relocate([@contig.clone.
+                   inserts(:open => 1, :close => 2, :sequence => 'TTT').
+                   inserts(:open => 7, :close => 8, :sequence => 'TTT').
+                   sequence('ATGTTTCCC')],
+                 [@record])
+      end
+
+      it{ should set_the_attribute(:seqname => 'scaffold') }
+      it{ should set_the_attribute(:phase   => 1) }
+      it{ should set_the_attribute(:strand  => '+') }
+
+      it{ should set_the_attribute(:start   => 5).only_for_the(:first) }
+      it{ should set_the_attribute(:end     => 7).only_for_the(:first) }
+
+    end
+
+    describe "reversed with an insert before an annotation" do
+
+      subject do
+        contig = @contig.clone.
+                 reverse(true).
+                 inserts(:open => 1, :close => 2, :sequence => 'TTT')
+        relocate([contig],[@record])
+      end
+
+      it{ should set_the_attribute(:seqname => 'scaffold') }
+      it{ should set_the_attribute(:phase   => 1) }
+      it{ should set_the_attribute(:strand  => '-') }
+
+      it{ should set_the_attribute(:start   => 1).only_for_the(:first) }
+      it{ should set_the_attribute(:end     => 3).only_for_the(:first) }
+
+    end
+
+    describe "with an insert overlapping with an annotation" do
+
+      subject do
+        relocate([@contig.clone.
+                   inserts(:open => 3, :close => 5, :sequence => 'TTT')],
+                 [@record])
+      end
+
+      it "should not include this annotation" do
+        subject.should be_empty
+      end
+
+    end
+
+    describe "with an annotation in a start trimmed region" do
+
+      subject do
+        relocate([@contig.clone.start(5)],[@record])
+      end
+
+      it "should not include this annotation" do
+        subject.should be_empty
+      end
+
+    end
+
+    describe "with an annotation in a stop trimmed region" do
+
+      subject do
+        relocate([@contig.clone.stop(5)],[@record])
+      end
+
+      it "should not include this annotation" do
+        subject.should be_empty
+      end
+
+    end
+
   end
 
   describe "relocating two contigs" do
@@ -203,16 +312,4 @@ describe Scaffolder::AnnotationLocator do
 
   end
 
-  describe "#flip_strand" do
-
-    it "should return '+' when passed '-'" do
-      described_class.flip_strand('+').should == '-'
-    end
-
-    it "should return '-' when passed '+'" do
-      described_class.flip_strand('-').should == '+'
-    end
-
-  end
-
 end

data/spec/scaffolder/gff_record_helper_spec.rb

@@ -0,0 +1,86 @@
+require File.expand_path(File.join(File.dirname(__FILE__), '..', 'spec_helper'))
+
+describe Scaffolder::GffRecordHelper do
+
+  it "should be included in Bio::GFF::GFF3::Record" do
+    Bio::GFF::GFF3::Record.ancestors.should include(described_class)
+  end
+
+  subject do
+    Bio::GFF::GFF3::Record.new(nil,nil,'CDS',1,3,nil,'+')
+  end
+
+  describe "#flip_strand" do
+
+    it "should change strand to '-' when flipped from '+'" do
+      subject.flip_strand
+      subject.strand.should == '-'
+    end
+
+    it "should change strand to '+' when flipped from '-'" do
+      subject.strand = '-'
+      subject.flip_strand
+      subject.strand.should == '+'
+    end
+
+  end
+
+  describe "#change_position_by" do
+
+    before do
+      subject.change_position_by 3
+    end
+
+    it "should increase start position" do
+      subject.start.should == 4
+    end
+
+    it "should increase end position" do
+      subject.end.should == 6
+    end
+
+  end
+
+  describe "#reverse_complement_by" do
+
+    before do
+      subject.reverse_complement_by 7
+    end
+
+    it "should increase start position" do
+      subject.start.should == 5
+    end
+
+    it "should increase end position" do
+      subject.end.should == 7
+    end
+
+    it "should flip the stand" do
+      subject.strand.should == '-'
+    end
+
+  end
+
+  describe "#overlap?" do
+
+    it "should return false when no overlap with a single insert" do
+      subject.overlap?(4..6).should be_false
+    end
+
+    it "should return true when overlapping with a single insert" do
+      subject.overlap?(0..1).should be_true
+      subject.overlap?(3..4).should be_true
+      subject.overlap?(2..4).should be_true
+    end
+
+    it "should return false when no overlap with a multiple inserts" do
+      subject.overlap?([4..6,7..9]).should be_false
+    end
+
+    it "should return true when overlapping with one of two inserts" do
+      subject.overlap?([0..1,4..6]).should be_true
+    end
+
+  end
+
+end

data/spec/spec_helper.rb

@@ -22,8 +22,8 @@ RSpec.configure do |config|
        a[:end], nil, a[:strand],  a[:phase])
     end
 
-    tmp = Tempfile.new("gff").path
-    File.open(tmp,'w'){ |out| out.print(gff) }
+    tmp = Tempfile.new("gff")
+    File.open(tmp.path,'w'){ |out| out.print(gff) }
     tmp
   end
 

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: scaffolder-annotation-locator
 version: !ruby/object:Gem::Version 
-  hash: 29
+  hash: 27
   prerelease: false
   segments: 
   - 0
-  - 0
   - 1
-  version: 0.0.1
+  - 0
+  version: 0.1.0
 platform: ruby
 authors: 
 - Michael Barton
@@ -15,20 +15,21 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-04-05 00:00:00 -04:00
+date: 2011-06-10 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
   requirement: &id001 !ruby/object:Gem::Requirement 
     none: false
     requirements: 
-    - - ~>
+    - - "="
       - !ruby/object:Gem::Version 
-        hash: 3
+        hash: 9
         segments: 
         - 0
         - 4
-        version: "0.4"
+        - 3
+        version: 0.4.3
   type: :runtime
   name: scaffolder
   prerelease: false
@@ -87,9 +88,9 @@ dependencies:
         hash: 19
         segments: 
         - 0
-        - 2
-        - 2
-        version: 0.2.2
+        - 3
+        - 0
+        version: 0.3.0
   type: :development
   name: scaffolder-test-helpers
   prerelease: false
@@ -155,12 +156,18 @@ files:
 - README.rdoc
 - Rakefile
 - VERSION
-- features/gff3.feature
+- features/inserts.feature
+- features/multiple-contigs.feature
+- features/single-contig.feature
 - features/step_definitions/scaffolder-annotation-locator_steps.rb
 - features/support/env.rb
+- features/unresolved.feature
 - lib/scaffolder/annotation_locator.rb
+- lib/scaffolder/extensions.rb
+- lib/scaffolder/gff_record_helper.rb
 - scaffolder-annotation-locator.gemspec
 - spec/scaffolder/annotation_locator_spec.rb
+- spec/scaffolder/gff_record_helper_spec.rb
 - spec/spec_helper.rb
 - spec/support/gff_attribute_matcher.rb
 has_rdoc: true
@@ -197,7 +204,5 @@ rubygems_version: 1.3.7
 signing_key: 
 specification_version: 3
 summary: Update locations of gff3 annotations from a scaffolder template
-test_files: 
-- spec/scaffolder/annotation_locator_spec.rb
-- spec/spec_helper.rb
-- spec/support/gff_attribute_matcher.rb
+test_files: []
+