scaffolder-tools 0.1.2 → 0.1.3

data/lib/scaffolder/tool/sequence.rb

@@ -17,18 +17,20 @@ class Scaffolder::Tool::Sequence < Scaffolder::Tool
   private
 
   def sequence(scaffold)
-    sequence = scaffold.inject(String.new) do |string,entry|
-      string << entry.sequence
-    end
+    scaffold.map{|entry| entry.sequence}.join
   end
 
   def header(sequence,opts={})
-    header = String.new
-    header << opts[:definition] + " " if opts[:definition]
-    unless opts[:no] && opts[:no][:sequence] && opts[:no][:sequence][:hash]
-      header << Digest::SHA1.hexdigest(sequence)
+    header = Array.new
+    header << opts[:definition] if opts[:definition]
+    if opts[:'with-sequence-digest']
+      digest = Digest::SHA1.hexdigest(sequence)
+      header << "[sha1=#{digest}]"
     end
-    header
+
+
+
+    header * ' '
   end
 
 end

data/lib/scaffolder/tool_index.rb

@@ -19,13 +19,6 @@ module Scaffolder::ToolIndex
     end
   end
 
-  def determine_tool(settings)
-    type = settings.rest.shift
-    tool_class = self[type]
-    settings[:unknown_tool] = type unless (tool_exists?(type) or type.nil?)
-    [tool_class,settings]
-  end
-
   private
 
   def tool_classes

data/man/scaffolder-format.7.ronn

@@ -0,0 +1,127 @@
+scaffolder-format(7) -- scaffolder authoring format using YAML
+==============================================================
+
+## SYNOPSIS
+
+The scaffolder software allows the construction of genome scaffolds using text
+files written in the YAML format. An example scaffold file with three entries
+is as follows:
+
+    ---
+    -
+      sequence:
+        source: 'sequence1'
+    -
+      unresolved:
+        length: 20
+    -
+      sequence:
+        source: 'sequence2'
+        start: 3
+        stop: 100
+        reverse: true
+        inserts:
+        -
+          source: 'insert1'
+          open: 5
+          close: 20
+
+## DESCRIPTION
+
+Scaffold files are written using the YAML format. The scaffold file along with
+the fasta file for the corresponding nucleotide sequences is used by the
+scaffolder software to generate the scaffold super sequence.
+
+The scaffold file should begin with three dashes `---`. This identifies the
+start of a YAML formatted document. The scaffold file is composed of a list of
+entries where the sequence for each entries is joined together to create the
+scaffold sequence. Each scaffold entry should should have a '-' dash line to
+indicate the start of the entry. The first line of the entry should define the
+type of entry. The entry types allowed are outlined in the section below.
+
+## SCAFFOLD ENTRY TYPES
+
+### SEQUENCE
+
+Sequence entries are the main component of the scaffold and insert the
+nucleotide sequences from the fasta file into the scaffold. The `sequence` tag
+is used to specify a sequence entry in the scaffold. The following tags can be
+used to modify how the sequence is inserted into the scaffold:
+
+  * `source:` <sequence-name>:
+    **Required.** Specify the source nucleotide sequence to insert into the
+    scaffold. The value <sequence-name> should match the first space delimited
+    word from the fasta header of the required sequence.
+
+  * `reverse:` <true|false>:
+    **Optional.** Specify whether sequence should be reverse completed in the
+    scaffold. Default value is false.
+
+  * `start:` <trim-position>:
+    **Optional.** Specify if the sequence should be trimmed from the start to
+    the specified coordinate position. Default value is 1 where no sequence is
+    removed.
+
+  * `stop:` <trim-position>:
+    **Optional.** Specify if the sequence should be trimmed from the end of the
+    sequence. Sequence onward of specified trim position is removed. Default
+    value is the length of the fasta sequence where no sequence is removed.
+
+  * `inserts:` <insert-list>:
+    **Optional.** Specify if additional insert sequences should be added to
+    replace regions in this sequence. Should be specified as a list of YAML
+    formatted [INSERT][] entries. Default value is an empty list.
+
+### UNRESOLVED
+
+Unresolved regions add sequences of N nucleotide characters to the scaffold.
+They define gaps in the scaffold and join separate sequence sequence regions
+together. Unresolved regions are specified using the `unresolved` tag. A single
+attribute tag is used to specify the length of an unresolved region.
+
+  * `length:` <length>:
+    **Required.** Specify the length of the unresolved region to add to the
+    scaffold.
+
+### INSERTS
+
+Insert sequences are used to replace regions in encoding [SEQUENCE][] entries.
+Each insert should correspond to a sequence in the fasta file. Inserts share
+the same attributes tags as [SEQUENCE][] entries but define two additional
+`open` and `close` attribute tags. These specify where the insert should be
+added to the encoding sequence. Multiple inserts can be added as a list to the
+encoding sequence entry using the `inserts` attribute tag. The following
+attribute tags can be to modify how an insert is added to the scaffold.
+
+  * `source:` <sequence-name>:
+    **Required.** Specify the insert sequence to use. The value <sequence-name>
+    should match the first space delimited word from the fasta header of the
+    required insert sequence.
+
+  * `reverse:` <true|false>:
+    **Optional.** Specify whether the insert should be reverse complemented.
+    Default value is false.
+
+  * `start:` <trim-position>:
+    **Optional.** Specify if the insert should be trimmed from the start to the
+    specified trim position. Default value is 1 where no sequence is removed.
+
+  * `stop:` <trim-position>:
+    **Optional.** Specify if the insert should be trimmed from the end of the
+    sequence. Sequence onward of trim position is removed. Default value is the
+    length of the fasta sequence where no sequence is removed.
+
+  * `open:` <insert-start-coordinate>:
+    **Optional if close coordinate is specified.** Specify the start coordinate
+    where the insert is added to the encoding sequence. Default value is the
+    close coordinate position minus the length of the insert sequence.
+
+  * `close:` <insert-close-coordinate>:
+    **Optional if open coordinate is specified.** Specify the end coordinate
+    position for adding the insert to the encoding sequence. Default value is
+    the open coordinate position plus the length of the insert sequence.
+
+## COPYRIGHT
+
+**Scaffolder** is Copyright (C) 2010 Michael Barton
+<http://michaelbarton.me.uk>

data/man/scaffolder-sequence.1.ronn

@@ -23,9 +23,9 @@ each contig specified in the scaffold file.
     Fasta definition line added at the top of the assembled sequence. The
     default is to use the SHA1 hash of the sequence as the definition.
 
-  * `--no-sequence-hash`:
-    Do not append the SHA1 hash of the assembled sequence to the definition
-    line.
+  * `--with-sequence-digest`:
+     Add an SHA1 digest of the assembled sequence to the definition line. This
+     may be useful for comparing large scaffold assemblies for identity.
 
 ## EXAMPLES
 
@@ -36,8 +36,6 @@ Pipe the assembly into another file.
 
 ## BUGS
 
-## BUGS
-
 **Scaffolder-sequence** is written in Ruby and depends on the scaffolder,
 BioRuby, configliere and ronn gems. See the Gemfile in the scaffolder-tools gem
 install directory for version details.

data/scaffolder-tools.gemspec

@@ -5,15 +5,14 @@
 
 Gem::Specification.new do |s|
   s.name = %q{scaffolder-tools}
-  s.version = "0.1.2"
+  s.version = "0.1.3"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
-  s.authors = ["Michael Barton"]
-  s.date = %q{2011-01-28}
-  s.default_executable = %q{scaffolder}
+  s.authors = [%q{Michael Barton}]
+  s.date = %q{2011-08-13}
   s.description = %q{Binary to use with scaffolder genome scaffolds}
-  s.email = %q{mail@michaelbarton.me.uk}
-  s.executables = ["scaffolder"]
+  s.email = %q{mail@next.gs}
+  s.executables = [%q{scaffolder}]
   s.extra_rdoc_files = [
     "LICENSE.txt",
     "README.rdoc"
@@ -21,12 +20,14 @@ Gem::Specification.new do |s|
   s.files = [
     ".document",
     ".rspec",
+    ".travis.yml",
     "Gemfile",
     "LICENSE.txt",
     "README.rdoc",
     "Rakefile",
     "VERSION",
     "bin/scaffolder",
+    "features/error_checking.feature",
     "features/help.feature",
     "features/sequence.feature",
     "features/step_definitions/scaffolder-tools.rb",
@@ -38,6 +39,7 @@ Gem::Specification.new do |s|
     "lib/scaffolder/tool/sequence.rb",
     "lib/scaffolder/tool/validate.rb",
     "lib/scaffolder/tool_index.rb",
+    "man/scaffolder-format.7.ronn",
     "man/scaffolder-help.1.ronn",
     "man/scaffolder-sequence.1.ronn",
     "man/scaffolder-validate.1.ronn",
@@ -51,24 +53,14 @@ Gem::Specification.new do |s|
     "spec/spec_helper.rb",
     "spec/support/exit_code_matcher.rb"
   ]
-  s.homepage = %q{http://www.michaelbarton.me.uk/projects/scaffolder/}
-  s.licenses = ["MIT"]
-  s.require_paths = ["lib"]
-  s.rubygems_version = %q{1.3.7}
+  s.homepage = %q{http://next.gs}
+  s.licenses = [%q{MIT}]
+  s.require_paths = [%q{lib}]
+  s.rubygems_version = %q{1.8.6}
   s.summary = %q{Tools for manipulating genome scaffolds}
-  s.test_files = [
-    "spec/scaffolder/binary_helper_spec.rb",
-    "spec/scaffolder/tool/help_spec.rb",
-    "spec/scaffolder/tool/sequence_spec.rb",
-    "spec/scaffolder/tool/validate_spec.rb",
-    "spec/scaffolder/tool_index_spec.rb",
-    "spec/scaffolder/tool_spec.rb",
-    "spec/spec_helper.rb",
-    "spec/support/exit_code_matcher.rb"
-  ]
+  s.test_files = [%q{spec/scaffolder/binary_helper_spec.rb}, %q{spec/scaffolder/tool/help_spec.rb}, %q{spec/scaffolder/tool/sequence_spec.rb}, %q{spec/scaffolder/tool/validate_spec.rb}, %q{spec/scaffolder/tool_index_spec.rb}, %q{spec/scaffolder/tool_spec.rb}, %q{spec/spec_helper.rb}, %q{spec/support/exit_code_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
@@ -78,15 +70,13 @@ Gem::Specification.new do |s|
       s.add_runtime_dependency(%q<ronn>, ["~> 0.7"])
       s.add_development_dependency(%q<bundler>, ["~> 1.0"])
       s.add_development_dependency(%q<jeweler>, ["~> 1.5"])
-      s.add_development_dependency(%q<gherkin>, ["~> 2.3.3"])
       s.add_development_dependency(%q<rspec>, ["~> 2.4"])
       s.add_development_dependency(%q<cucumber>, ["~> 0.10"])
       s.add_development_dependency(%q<fakefs>, ["~> 0.2"])
       s.add_development_dependency(%q<aruba>, ["~> 0.2"])
       s.add_development_dependency(%q<mocha>, ["~> 0.9"])
-      s.add_development_dependency(%q<hashie>, ["~> 0.4"])
       s.add_development_dependency(%q<yard>, ["~> 0.6"])
-      s.add_development_dependency(%q<scaffolder-test-helpers>, ["~> 0.1"])
+      s.add_development_dependency(%q<scaffolder-test-helpers>, ["~> 0.4"])
     else
       s.add_dependency(%q<configliere>, ["~> 0.1"])
       s.add_dependency(%q<bio>, ["~> 1.4"])
@@ -94,15 +84,13 @@ Gem::Specification.new do |s|
       s.add_dependency(%q<ronn>, ["~> 0.7"])
       s.add_dependency(%q<bundler>, ["~> 1.0"])
       s.add_dependency(%q<jeweler>, ["~> 1.5"])
-      s.add_dependency(%q<gherkin>, ["~> 2.3.3"])
       s.add_dependency(%q<rspec>, ["~> 2.4"])
       s.add_dependency(%q<cucumber>, ["~> 0.10"])
       s.add_dependency(%q<fakefs>, ["~> 0.2"])
       s.add_dependency(%q<aruba>, ["~> 0.2"])
       s.add_dependency(%q<mocha>, ["~> 0.9"])
-      s.add_dependency(%q<hashie>, ["~> 0.4"])
       s.add_dependency(%q<yard>, ["~> 0.6"])
-      s.add_dependency(%q<scaffolder-test-helpers>, ["~> 0.1"])
+      s.add_dependency(%q<scaffolder-test-helpers>, ["~> 0.4"])
     end
   else
     s.add_dependency(%q<configliere>, ["~> 0.1"])
@@ -111,15 +99,13 @@ Gem::Specification.new do |s|
     s.add_dependency(%q<ronn>, ["~> 0.7"])
     s.add_dependency(%q<bundler>, ["~> 1.0"])
     s.add_dependency(%q<jeweler>, ["~> 1.5"])
-    s.add_dependency(%q<gherkin>, ["~> 2.3.3"])
     s.add_dependency(%q<rspec>, ["~> 2.4"])
     s.add_dependency(%q<cucumber>, ["~> 0.10"])
     s.add_dependency(%q<fakefs>, ["~> 0.2"])
     s.add_dependency(%q<aruba>, ["~> 0.2"])
     s.add_dependency(%q<mocha>, ["~> 0.9"])
-    s.add_dependency(%q<hashie>, ["~> 0.4"])
     s.add_dependency(%q<yard>, ["~> 0.6"])
-    s.add_dependency(%q<scaffolder-test-helpers>, ["~> 0.1"])
+    s.add_dependency(%q<scaffolder-test-helpers>, ["~> 0.4"])
   end
 end
 

data/spec/scaffolder/binary_helper_spec.rb

@@ -1,4 +1,4 @@
-require File.join(File.dirname(__FILE__),'..','spec_helper')
+require 'spec_helper'
 
 describe Scaffolder::BinaryHelper do
 

data/spec/scaffolder/tool/help_spec.rb

@@ -1,4 +1,4 @@
-require File.expand_path(File.join(File.dirname(__FILE__),'..','..','spec_helper'))
+require 'spec_helper'
 
 describe Scaffolder::Tool::Help do
 
@@ -18,108 +18,98 @@ describe Scaffolder::Tool::Help do
       Commands:
     MSG
 
-  before(:each) do
-    @settings = Hash.new
-    @settings.stubs(:rest).returns([])
-  end
-
-  describe "execution with no command and the version argument" do
+  describe "execution with" do
 
-    subject do
-      @settings[:version] = true
-      @settings[:empty_args] = true
-      described_class.new(@settings)
+    before(:each) do
+      @settings = Hash.new
+      @settings.stubs(:rest).returns([])
     end
 
-    it "should not raise an error" do
-      lambda{ subject.execute }.should_not raise_error
-    end
+    describe "unknown command" do
 
-    it "should return the version number" do
-      version = File.read('VERSION').strip
-      subject.execute.should == "scaffolder tool version " + version
-    end
-
-  end
+      subject do
+        @settings[:unknown_tool] = 'anything'
+        described_class.new(@settings)
+      end
 
-  describe "execution with no command" do
+      it "should raise an error" do
+        lambda{ subject.execute }.should(raise_error(ArgumentError,
+          "Unknown command 'anything'.\nSee 'scaffolder help'."))
+      end
 
-    subject do
-      @settings[:empty_args] = true
-      described_class.new(@settings)
     end
 
-    it "should not raise an error" do
-      lambda{ subject.execute }.should_not raise_error
-    end
+    describe "no arguments" do
 
-    it "should contain the usage information" do
-      subject.execute.should include(USAGE)
-    end
+      subject do
+        @settings[:empty_args] = true
+        described_class.new(@settings)
+      end
 
-    it "should contain each tool information" do
-      tool_subclasses.each do |cls|
-        subject.execute.should include(cls.description)
+      it "should not raise an error" do
+        lambda{ subject.execute }.should_not raise_error
       end
-    end
 
-  end
+      it "should contain the usage information" do
+        subject.execute.should include(USAGE)
+      end
 
-  describe "execution with an invalid command arguments" do
+      it "should contain each tool information" do
+        tool_subclasses.each do |cls|
+          subject.execute.should include(cls.description)
+        end
+      end
 
-    subject do
-      @settings[:unknown_tool] = 'unknown_command'
-      described_class.new(@settings)
     end
 
-    it "should raise an error" do
-      lambda{ subject.execute }.should(raise_error(ArgumentError,
-        "Unknown command 'unknown_command'.\nSee 'scaffolder help'."))
-    end
+    describe "version argument" do
 
-  end
+      subject do
+        @settings[:version] = true
+        @settings[:empty_args] = true
+        described_class.new(@settings)
+      end
 
-  describe "execution with the name of a scaffolder tool command" do
+      it "should not raise an error" do
+        lambda{ subject.execute }.should_not raise_error
+      end
 
-    before(:each) do
-      @tool = Class.new(Scaffolder::Tool)
-      described_class.superclass.const_set('Fake',@tool)
+      it "should return the version number" do
+        version = File.read('VERSION').strip
+        subject.execute.should == "scaffolder tool version " + version
+      end
 
-      @man_dir = File.join(%W|#{File.dirname(__FILE__)} .. .. .. man| )
-      @fake_man = File.join(@man_dir,'scaffolder-fake.1.ronn')
     end
 
-    after(:each) do
-      described_class.superclass.send(:remove_const,'Fake')
-    end
+    describe "existing man page" do
 
-    subject do
-      @settings.stubs(:rest).returns(['fake'])
-      described_class.new(@settings)
-    end
+      before(:each) do
+        @tool = Class.new(Scaffolder::Tool)
+        described_class.superclass.const_set('Fake',@tool)
 
-    it "should not raise an error" do
-      Kernel.stubs(:system).with("ronn -m #{File.expand_path(@fake_man)}")
-      lambda{ subject.execute }.should_not raise_error
-    end
+        @man_dir = File.join(%W|#{File.dirname(__FILE__)} .. .. .. man| )
+        @fake_man = File.join(@man_dir,'scaffolder-fake.1.ronn')
+      end
 
-    it "should call ronn on the command line with the man file location" do
-      Kernel.expects(:system).with("ronn -m #{File.expand_path(@fake_man)}")
-      subject.execute
-    end
+      after(:each) do
+        described_class.superclass.send(:remove_const,'Fake')
+      end
 
-  end
+      subject do
+        @settings.stubs(:rest).returns(['fake'])
+        described_class.new(@settings)
+      end
 
-  describe "execution with the name of a unknown command" do
+      it "should not raise an error" do
+        Kernel.stubs(:system).with("ronn -m #{File.expand_path(@fake_man)}")
+        lambda{ subject.execute }.should_not raise_error
+      end
 
-    subject do
-      @settings.stubs(:rest).returns(['fake'])
-      described_class.new(@settings)
-    end
+      it "should call ronn on the command line with the man file location" do
+        Kernel.expects(:system).with("ronn -m #{File.expand_path(@fake_man)}")
+        subject.execute
+      end
 
-    it "should raise an error" do
-      lambda{ subject.execute }.should(raise_error(ArgumentError,
-        "Unknown command 'fake'.\nSee 'scaffolder help'."))
     end
 
   end