genomer-plugin-validate 0.0.1

data/lib/genomer-plugin-validate/validator/view_attributes.rb

@@ -0,0 +1,16 @@
+class GenomerPluginValidate::Validator::ViewAttributes < Genomer::Plugin
+
+  def valid_view_attributes
+    %w|product ec_number function feature_type|
+  end
+
+  def run
+    annotations.
+      map{|attn| attn.attributes.map{|(k,v)| [k,attn] }}.
+      flatten(1).
+      select{|(term,_)| term =~ (/^[a-z]/) }.
+      reject{|(term,_)| valid_view_attributes.include? term }.
+      map{|(term,attn)| "Illegal view attribute '#{term}' for '#{attn.id}'"}
+  end
+
+end

data/man/genomer-validate.ronn

@@ -0,0 +1,100 @@
+genomer-validate(1) -- Test assembly files for errors
+=====================================================
+
+## SYNOPSIS
+
+`genomer validate` <type>
+
+## DESCRIPTION
+
+**Genomer-validate** tests the assembly files for errors. Any detected errors 
+are reported to the terminal.
+
+## OPTIONS
+
+  * `--no_view_validations`
+    The GFF ninth-column attributes are validated to ensure that all attributes 
+    beginning with a lower case character match those used by 
+    genomer-plugin-view to generate GenBank annotation table output. This 
+    argument can be used to disable these validations.
+
+## FILE TYPES
+
+### ANNOTATIONS
+
+Test the annotation file for errors. The default location for this file is 
+**assembly/annotations.gff**. The following validations are performed:
+
+  * `Identical locations`:
+    Flag annotations of the same type with identical start and end coordinates.
+
+  * `Identical IDs`:
+    Flag two or more annotations identical ID attributes.
+
+  * `Missing ID`:
+    Flag annotations without an ID attribute.
+
+  * `Missing either Name or product`:
+    Flag annotations without either a Name or product attribute.
+
+  * `Capitalised Name attribute`:
+    Flag annotations with Name attributes beginning with an uppercase letter.
+
+  * `Invalid GFF attributes`:
+    Flag annotations with capitalised attribute keys outside of the GFF3 
+    vocabulary. See below for a description of this vocabulary.
+
+  * `Invalid genomer plugin view attributes`:
+    Flag annotations with attribute keys outside of the genomer-plugin-view 
+    vocabulary. This validation can be disabled using the `--validate_for_view` 
+    command line flag.
+
+  * `Bad product fields`: NCBI requires that annotation product fields not
+  contain terms such as "putative" or end with "-like." This validator checks
+  that product fields do not contain the more common types of disallowed terms. 
+
+The [GFF3 specification][http://www.sequenceontology.org/gff3.shtml] restricts 
+attributes beginning with an upper case character (A-Z) in the ninth column to 
+a controlled vocabulary. Any attribute keys appearing in the ninth column not 
+included in this vocabulary will raise a validation error. The allowed keys are 
+listed below. The ID field is considered mandatory in this software and will 
+raise an error if not present.
+
+  * ID
+  * Name
+  * Note
+  * Alias
+  * Parent
+  * Target
+  * Gap
+  * Derives\_from
+  * Dbxref
+  * Ontology\_term
+  * Is\_circular
+
+Any fields beginning with a lower case character (a-z) are allowed in the GFF3 
+ninth column. The genomer-plugin-view however uses these lower case fields to 
+populate a GenBank annotation table and a validation is performed for this.
+The optional `--no_view_validations` flag can be used to disable this 
+validation. The allowed genomer-plugin-view attribute keys are as follows:
+
+  * product
+  * ec\_number
+  * function
+  * feature\_type
+
+## EXAMPLES
+
+Test the annotation file for errors.
+
+    $ genomer validate annotations
+
+## BUGS
+
+**Genomer-validate** is written in Ruby and depends on the genomer gem. See the 
+Gemfile in the genomer-plugin-validate gem install directory for version 
+details.
+
+## COPYRIGHT
+
+**Genomer** is Copyright (C) 2012 Michael Barton <http://michaelbarton.me.uk>

data/spec/genomer-plugin-validate/group/annotations_spec.rb

@@ -0,0 +1,18 @@
+require 'spec_helper'
+require 'genomer-plugin-validate/group/annotations'
+
+describe GenomerPluginValidate::Group::Annotations do
+
+  its("class.description"){ should == "Validate GFF3 annotations file" }
+
+  its("class.validators"){ should be_instance_of Array }
+  its("class.validators"){ should include :duplicate_id }
+  its("class.validators"){ should include :missing_id }
+  its("class.validators"){ should include :no_name_or_product }
+  its("class.validators"){ should include :gff3_attributes }
+  its("class.validators"){ should include :view_attributes }
+  its("class.validators"){ should include :duplicate_coordinates }
+  its("class.validators"){ should include :uppercase_name }
+  its("class.validators"){ should include :bad_product_field }
+
+end

data/spec/genomer-plugin-validate/group_spec.rb

@@ -0,0 +1,24 @@
+require 'spec_helper'
+
+describe GenomerPluginValidate::Group do
+
+  before do
+    @example = GenomerPluginValidate::Group::Example = Class.new
+    stub(described_class).require(anything)
+  end
+
+  after do
+    GenomerPluginValidate::Group.send(:remove_const,'Example')
+  end
+
+  describe "#groups" do
+
+    subject do
+      described_class.groups
+    end
+
+    its(['example']){should ==  @example}
+
+  end
+
+end

data/spec/genomer-plugin-validate/validator/bad_product_field_spec.rb

@@ -0,0 +1,93 @@
+require 'spec_helper'
+require 'genomer-plugin-validate/validator/bad_product_field'
+
+describe GenomerPluginValidate::Validator::BadProductField do
+
+  subject{ described_class }
+
+  describe "where there are no annotations" do
+    it{ should return_no_errors_for [] }
+  end
+
+  describe "where there is valid annotation product field" do
+    it{ should return_no_errors_for [annotation_with_product("transmembrane protein")]}
+  end
+
+  describe "where the product field begins with hypothetical" do
+    it do
+      should return_errors_for(
+        [annotation_with_product("hypothetical something")],
+        ["Bad product field for '1:' start with 'putative' instead of 'hypothetical.'"]
+      )
+    end
+
+    it do
+      should return_errors_for(
+        [annotation_with_product("Hypothetical somthing")],
+        ["Bad product field for '1:' start with 'putative' instead of 'hypothetical.'"]
+      )
+    end
+  end
+
+  describe "where the product field contains only 'hypothetical protein'" do
+    it{ should return_no_errors_for [annotation_with_product("hypothetical protein")]}
+    it{ should return_no_errors_for [annotation_with_product("Hypothetical protein.")]}
+  end
+
+  describe "where the product ends with with like/domain/related" do
+    start = "Bad product field for '1:' "
+    ['like','related','domain'].each do |word|
+      [word, word.capitalize].each do |cased|
+        [' '+cased, '-'+cased].each do |prefixed|
+          [prefixed+'.', prefixed].each do |formatted|
+            it do
+              should return_errors_for(
+                [annotation_with_product("membrane" + formatted)],
+                [start + "products ending with '#{word}' are not allowed."]
+              )
+            end
+          end
+        end
+      end
+    end
+  end
+
+  describe "where the product field contains n-term" do
+    it do
+      should return_errors_for(
+        [annotation_with_product("n-term membrane region")],
+        ["Bad product field for '1:' 'N-terminal' or variations are not allowed."]
+      )
+    end
+
+    it do
+      should return_errors_for(
+        [annotation_with_product("N terminal membrane region")],
+        ["Bad product field for '1:' 'N-terminal' or variations are not allowed."]
+      )
+    end
+  end
+
+  describe "where the product field contains similiar to 'n terminal'" do
+    it do
+      should return_no_errors_for [annotation_with_product("transcription termination")]
+    end
+  end
+
+  describe "where the product field is in all caps" do
+    it do
+      should return_errors_for(
+        [annotation_with_product("PROTEIN")],
+        ["Bad product field for '1:' all caps product fields are not allowed."]
+      )
+    end
+
+    it do
+      should return_errors_for(
+        [annotation_with_product("MEMBRANE PROTEIN")],
+        ["Bad product field for '1:' all caps product fields are not allowed."]
+      )
+    end
+  end
+
+end

data/spec/genomer-plugin-validate/validator/duplicate_coordinates_spec.rb

@@ -0,0 +1,24 @@
+require 'spec_helper'
+require 'genomer-plugin-validate/validator/duplicate_coordinates'
+
+describe GenomerPluginValidate::Validator::DuplicateCoordinates do
+
+  subject{ described_class }
+
+  describe "where there are no annotations" do
+    it{ should return_no_errors_for [] }
+  end
+
+  describe "where there are two annotations with different coordinates" do
+    attns = [annotation(:start => 1, :end => 3), annotation(:start => 4, :end => 6)]
+    it{ should return_no_errors_for attns}
+  end
+
+  describe "where there are two annotations with the same coordinates" do
+    attns = [annotation(:start => 1, :end => 3, :attributes => {'ID' => '1'}),
+             annotation(:start => 1, :end => 3, :attributes => {'ID' => '2'})]
+    errors = ["Identical locations for '1', '2'"]
+    it{ should return_errors_for attns, errors}
+  end
+
+end

data/spec/genomer-plugin-validate/validator/duplicate_id_spec.rb

@@ -0,0 +1,34 @@
+require 'spec_helper'
+require 'genomer-plugin-validate/validator/duplicate_id'
+
+describe GenomerPluginValidate::Validator::DuplicateID do
+
+  subject{ described_class }
+
+  describe "where there are no annotations" do
+    it{ should return_no_errors_for [] }
+  end
+
+  describe "where there are two annotations with different IDs" do
+    it{ should return_no_errors_for [annotation_with_id(1), annotation_with_id(2)]}
+  end
+
+  describe "where there two annotations with missing IDs" do
+    attns = [Annotation.new.to_gff3_record, Annotation.new.to_gff3_record]
+    it{ should return_no_errors_for attns}
+  end
+
+  describe "where there are two annotations with the same ID" do
+    attns  = [annotation_with_id(1), annotation_with_id(1)]
+    errors = ["Duplicate ID '1'"]
+    it{ should return_errors_for attns, errors}
+  end
+
+  describe "where there are two sets of annotations with the same ID" do
+    attns  = [annotation_with_id(1), annotation_with_id(1),
+              annotation_with_id(2), annotation_with_id(2)]
+    errors = ["Duplicate ID '1'", "Duplicate ID '2'"]
+    it{ should return_errors_for attns, errors}
+  end
+
+end

data/spec/genomer-plugin-validate/validator/gff_attributes_spec.rb

@@ -0,0 +1,32 @@
+require 'spec_helper'
+require 'genomer-plugin-validate/validator/gff3_attributes'
+
+describe GenomerPluginValidate::Validator::Gff3Attributes do
+
+  subject{ described_class }
+
+  describe "where there are no annotations" do
+    it{ should return_no_errors_for [] }
+  end
+
+  describe "where there are valid GFF3 attributes" do
+    attrs = %w|ID Name Alias Parent Target Gap Derives_from
+               Note Dbxref Ontology_term Is_circular|
+    attrs.each do |attr|
+      attns = [annotation({:attributes => {attr => 'something'}})]
+      it{ should return_no_errors_for attns}
+    end
+  end
+
+  describe "where there are lower case attribute keys" do
+    attns = [annotation({:attributes => {'unknown_term' => 'something'}})]
+    it{ should return_no_errors_for attns}
+  end
+
+  describe "where there is an unknown capitalised attribute key" do
+    attns =  [annotation({:attributes => {'Unknown_term' => 'something','ID' => 1}})]
+    errors = ["Illegal GFF3 attribute 'Unknown_term' for '1'"]
+    it{ should return_errors_for attns, errors}
+  end
+
+end

data/spec/genomer-plugin-validate/validator/missing_id_spec.rb

@@ -0,0 +1,27 @@
+require 'spec_helper'
+require 'genomer-plugin-validate/validator/missing_id'
+
+describe GenomerPluginValidate::Validator::MissingID do
+
+  subject{ described_class }
+
+  describe "where there are no annotations" do
+    it{ should return_no_errors_for [] }
+  end
+
+  describe "where there no annotations with missing IDs" do
+    it{ should return_no_errors_for [annotation_with_id(1), annotation_with_id(2)] }
+  end
+
+  describe "where there an annotation with a missing ID" do
+    errors = ["Annotations found with missing ID attribute"]
+    it{ should return_errors_for [Annotation.new.to_gff3_record], errors}
+  end
+
+  describe "where there are multiple annotations with missing IDs" do
+    errors = ["Annotations found with missing ID attribute"]
+    attns  = [Annotation.new.to_gff3_record, Annotation.new.to_gff3_record]
+    it{ should return_errors_for attns, errors}
+  end
+
+end

data/spec/genomer-plugin-validate/validator/no_name_or_product_spec.rb

@@ -0,0 +1,28 @@
+require 'spec_helper'
+require 'genomer-plugin-validate/validator/no_name_or_product'
+
+describe GenomerPluginValidate::Validator::NoNameOrProduct do
+
+  subject{ described_class }
+
+  describe "where there are no annotations" do
+    it{ should return_no_errors_for [] }
+  end
+
+  describe "where an annotation has a Name attribute" do
+    attns = [annotation({:attributes => {'Name' => 'something'}})]
+    it{ should return_no_errors_for attns}
+  end
+
+  describe "where an annotation has a Product attribute" do
+    attns = [annotation({:attributes => {'product' => 'something'}})]
+    it{ should return_no_errors_for attns}
+  end
+
+  describe "where an annotation has neither a Name nor a product attribute" do
+    attns  = [annotation({:attributes => {'ID' => '1'}})]
+    errors = ["No 'Name' or 'product' attribute for annotation '1'"]
+    it{ should return_errors_for attns, errors}
+  end
+
+end

data/spec/genomer-plugin-validate/validator/uppercase_name_spec.rb

@@ -0,0 +1,22 @@
+require 'spec_helper'
+require 'genomer-plugin-validate/validator/uppercase_name'
+
+describe GenomerPluginValidate::Validator::UppercaseName do
+
+  subject{ described_class }
+
+  describe "where there are no annotations" do
+    it{ should return_no_errors_for [] }
+  end
+
+  describe "where there is an annotations with a lowercase name" do
+    it{ should return_no_errors_for [annotation(:attributes => {'Name' => 'something'})]}
+  end
+
+  describe "where there two annotations with an uppercase name" do
+    attns = [annotation(:attributes => {'Name' => 'Something', 'ID' => 1})]
+    errors = ["Illegal capitalised Name attribute 'Something' for '1'"]
+    it{ should return_errors_for attns, errors}
+  end
+
+end

data/spec/genomer-plugin-validate/validator/view_attributes_spec.rb

@@ -0,0 +1,31 @@
+require 'spec_helper'
+require 'genomer-plugin-validate/validator/view_attributes'
+
+describe GenomerPluginValidate::Validator::ViewAttributes do
+
+  subject{ described_class }
+
+  describe "where there are no annotations" do
+    it{ should return_no_errors_for [] }
+  end
+
+  describe "where there are capitalized attribute keys" do
+    attns = [annotation({:attributes => {'Unknown_term' => 'something'}})]
+    it{ should return_no_errors_for attns}
+  end
+
+  describe "where there are valid view attributes" do
+    attrs = %w|product ec_number function feature_type|
+    attrs.each do |attr|
+      attns = [annotation({:attributes => {attr => 'something'}})]
+      it{ should return_no_errors_for attns}
+    end
+  end
+
+  describe "where there is an unknown lower case attribute key" do
+    attns =  [annotation({:attributes => {'unknown_term' => 'something','ID' => 1}})]
+    errors = ["Illegal view attribute 'unknown_term' for '1'"]
+    it{ should return_errors_for attns, errors}
+  end
+
+end