scaffolder-tools 0.1.0

data/lib/scaffolder/tool/help.rb

@@ -0,0 +1,57 @@
+require 'scaffolder/tool'
+
+class Scaffolder::Tool::Help < Scaffolder::Tool
+  include Scaffolder::ToolIndex
+
+  def self.description
+    "Help information for scaffolder commands"
+  end
+
+  def execute
+    raise_for_unknown(@settings[:unknown_tool]) if @settings[:unknown_tool]
+
+    tool = settings.rest.first
+    if tool
+      raise_for_unknown(tool) unless tool_exists?(tool)
+      man settings.rest.first
+    elsif @settings[:version]
+      return version
+    else
+      return help
+    end
+  end
+
+  private
+
+  def raise_for_unknown(command)
+    msg = "Unknown command '#{command}'.\nSee 'scaffolder help'."
+    raise ArgumentError.new(msg)
+  end
+
+  def version
+    ver = File.read(File.join(%W|#{File.dirname(__FILE__)} .. .. .. VERSION|)).strip
+    "scaffolder tool version #{ver}"
+  end
+
+  def help
+    string = <<-MSG.gsub(/^ {6}/, '')
+      usage: scaffolder [--version] COMMAND scaffold-file sequence-file
+      [options]
+
+      Commands:
+    MSG
+    [:help,:sequence,:validate].each do |name|
+      string << "  "
+      string << name.to_s.ljust(12)
+      string << tools[name].description + "\n"
+    end
+    string
+  end
+
+  def man(tool)
+    man_page = File.join(
+      %W|#{File.dirname(__FILE__)} .. .. .. man scaffolder-#{tool}.1.ronn|)
+
+    Kernel.system("ronn -m #{File.expand_path(man_page)}")
+  end
+end

data/lib/scaffolder/tool/sequence.rb

@@ -0,0 +1,34 @@
+require 'yaml'
+require 'digest/sha1'
+require 'scaffolder'
+require 'scaffolder/tool'
+
+class Scaffolder::Tool::Sequence < Scaffolder::Tool
+
+  def self.description
+    "Generate the fasta output for the scaffold"
+  end
+
+  def execute
+    s = sequence(scaffold)
+    Bio::Sequence.new(s).output(:fasta,:header => header(s,@settings))
+  end
+
+  private
+
+  def sequence(scaffold)
+    sequence = scaffold.inject(String.new) do |string,entry|
+      string << entry.sequence
+    end
+  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)
+    end
+    header
+  end
+
+end

data/lib/scaffolder/tool/validate.rb

@@ -0,0 +1,51 @@
+require 'yaml'
+require 'scaffolder'
+require 'scaffolder/tool'
+
+class Scaffolder::Tool::Validate < Scaffolder::Tool
+
+  def self.description
+    "Validate scaffold for overlapping inserts"
+  end
+
+  def execute
+    bad_sequences = errors
+    return if bad_sequences.empty?
+
+    output = bad_sequences.inject(Array.new) do |array, sequence|
+      self.class.sequence_errors(sequence).each do |overlaps|
+        array << {'sequence-insert-overlap' => {
+          'source' => sequence.source,
+          'inserts' => overlaps.map do |overlap|
+            {'open'  => overlap.open,
+            'close'  => overlap.close,
+            'source' => overlap.source}
+          end
+        }}
+      end
+      array
+    end
+
+    YAML.dump(output)
+  end
+
+  def errors
+    sequences = scaffold.select{|i| i.entry_type == :sequence}
+    sequences.reject{|i| self.class.sequence_errors(i).empty? }
+  end
+
+  def self.inserts_overlap?(a,b)
+    ! (a.position.to_a & b.position.to_a).empty?
+  end
+
+  def self.sequence_errors(sequence)
+    sequence.inserts.inject(Array.new) do |errors,a|
+      sequence.inserts.each do |b|
+        next if a.equal?(b)
+        errors << [a,b].sort if inserts_overlap?(a,b)
+      end
+      errors
+    end.uniq
+  end
+
+end

data/lib/scaffolder/tool.rb

@@ -0,0 +1,44 @@
+require 'scaffolder'
+
+class Scaffolder::Tool
+  require 'scaffolder/tool_index'
+
+  attr :scaffold_file
+  attr :sequence_file
+  attr :settings
+
+  def initialize(settings)
+    @scaffold_file = settings.rest.first
+    @sequence_file = settings.rest.last
+    @settings = settings
+  end
+
+  def run(out=STDOUT,err=STDERR)
+    begin
+      message = execute
+    rescue Exception => e
+      err.puts("Error. #{e.message}")
+      exit(1)
+    else
+      out.puts(message) if message
+      exit(0)
+    end
+  end
+
+  def scaffold
+    {:Scaffold => @scaffold_file, :Sequence => @sequence_file}.each do |name,file|
+      unless File.exists?(file)
+        raise ArgumentError.new("#{name} file not found: #{file}")
+      end
+      if File.size(file) == 0
+        raise ArgumentError.new("#{name} file is empty: #{file}")
+      end
+    end
+
+    Scaffolder.new(YAML.load(File.read(@scaffold_file)),@sequence_file)
+  end
+
+  Dir["#{File.dirname(__FILE__)}/tool/*.rb"].each do |f|
+    require File.expand_path(f)
+  end
+end

data/lib/scaffolder/tool_index.rb

@@ -0,0 +1,50 @@
+require 'scaffolder'
+
+module Scaffolder::ToolIndex
+  require 'scaffolder/tool'
+
+  def get_tool(name)
+    tools[normalise(name)]
+  end
+
+  def tool_exists?(name)
+    ! get_tool(name).nil?
+  end
+
+  def [](type)
+    if tool_exists?(type)
+      get_tool(type)
+    else
+      Scaffolder::Tool::Help
+    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
+    Scaffolder::Tool.constants.inject(Array.new) do |array,constant|
+      clss = Scaffolder::Tool.const_get(constant)
+      array << clss if clss.superclass == Scaffolder::Tool
+      array
+    end
+  end
+
+  def tools
+    tool_classes.inject(Hash.new) do |hash,clss|
+      hash[clss.to_s.split('::').last.downcase.to_sym] = clss
+      hash
+    end
+  end
+
+  def normalise(name)
+    name.to_s.downcase.to_sym if name
+  end
+
+end

data/man/scaffolder-help.1.ronn

@@ -0,0 +1,23 @@
+scaffolder-help(1) -- display help information about scaffolder commands
+========================================================================
+
+## SYNOPSIS
+
+`scaffolder help` [COMMAND]
+
+## DESCRIPTION
+
+When run with no options **scaffolder-help** returns a list of available
+commands. When run with a scaffolder command name the corresponding man page is
+shown.
+
+## BUGS
+
+**Scaffolder-help** 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.
+
+## COPYRIGHT
+
+**Scaffolder** is Copyright (C) 2010 Michael Barton
+<http://michaelbarton.me.uk>

data/man/scaffolder-sequence.1.ronn

@@ -0,0 +1,48 @@
+scaffolder-sequence(1) -- assemble yaml scaffold file into fasta sequence
+=========================================================================
+
+## SYNOPSIS
+
+`scaffolder sequence` [<options>...] <scaffold-file> <sequence-file>
+
+## DESCRIPTION
+
+**Scaffolder-sequence** assembles a single fasta nucleotide sequence from the
+sequences and coordinates specified in the **scaffold-file**.
+
+## FILES
+
+**Scaffolder-sequence** requires two files to create a fasta assembly. The
+**scaffold-file** specifies the positions of individual contigs in the scaffold
+assembly. The **sequence-file** contains the corresponding fasta sequences of
+each contig specified in the scaffold file.
+
+## OPTIONS
+
+  * `--definition`=[<definition>]:
+    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.
+
+## EXAMPLES
+
+Assemble sequence using a scaffold file and fasta formatted set of contigs.
+Pipe the assembly into another file.
+
+    $ scaffolder-sequence scaffold.yml contigs.fna > assembly.fna
+
+## 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.
+
+## COPYRIGHT
+
+**Scaffolder** is Copyright (C) 2010 Michael Barton
+<http://michaelbarton.me.uk>

data/man/scaffolder-validate.1.ronn

@@ -0,0 +1,28 @@
+scaffolder-validate(1) -- Test scaffold for overlapping inserts
+===============================================================
+
+## SYNOPSIS
+
+`scaffolder validate` <scaffolder-file> <sequence-file>
+
+## DESCRIPTION
+
+**Scaffolder-validate** examines the **scaffolder-file** to detect overlapping
+sequence inserts. Overlapping sequence inserts may lead to unexpected sequence
+output in generated scaffold sequence and should be avoided.  A list of any
+overlapping inserts is returned in a yaml format.
+
+## EXAMPLES
+
+    $ scaffolder-validate scaffold.yml sequences.fna > overlapping_inserts.yml
+
+## BUGS
+
+**Scaffolder-validate** 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.
+
+## COPYRIGHT
+
+**Scaffolder** is Copyright (C) 2010 Michael Barton
+<http://michaelbarton.me.uk>

data/scaffolder-tools.gemspec

@@ -0,0 +1,125 @@
+# Generated by jeweler
+# DO NOT EDIT THIS FILE DIRECTLY
+# Instead, edit Jeweler::Tasks in Rakefile, and run 'rake gemspec'
+# -*- encoding: utf-8 -*-
+
+Gem::Specification.new do |s|
+  s.name = %q{scaffolder-tools}
+  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{2011-01-04}
+  s.default_executable = %q{scaffolder}
+  s.description = %q{Binary to use with scaffolder genome scaffolds}
+  s.email = %q{mail@michaelbarton.me.uk}
+  s.executables = ["scaffolder"]
+  s.extra_rdoc_files = [
+    "LICENSE.txt",
+    "README.rdoc"
+  ]
+  s.files = [
+    ".document",
+    ".rspec",
+    "Gemfile",
+    "LICENSE.txt",
+    "README.rdoc",
+    "Rakefile",
+    "VERSION",
+    "bin/scaffolder",
+    "features/help.feature",
+    "features/sequence.feature",
+    "features/step_definitions/scaffolder-tools.rb",
+    "features/support/env.rb",
+    "features/validate.feature",
+    "lib/scaffolder/binary_helper.rb",
+    "lib/scaffolder/tool.rb",
+    "lib/scaffolder/tool/help.rb",
+    "lib/scaffolder/tool/sequence.rb",
+    "lib/scaffolder/tool/validate.rb",
+    "lib/scaffolder/tool_index.rb",
+    "man/scaffolder-help.1.ronn",
+    "man/scaffolder-sequence.1.ronn",
+    "man/scaffolder-validate.1.ronn",
+    "scaffolder-tools.gemspec",
+    "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.homepage = %q{http://github.com/michaelbarton/scaffolder-tools}
+  s.licenses = ["MIT"]
+  s.require_paths = ["lib"]
+  s.rubygems_version = %q{1.3.7}
+  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"
+  ]
+
+  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<configliere>, ["~> 0.1"])
+      s.add_runtime_dependency(%q<bio>, ["~> 1.4"])
+      s.add_runtime_dependency(%q<scaffolder>, ["~> 0.4"])
+      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"])
+    else
+      s.add_dependency(%q<configliere>, ["~> 0.1"])
+      s.add_dependency(%q<bio>, ["~> 1.4"])
+      s.add_dependency(%q<scaffolder>, ["~> 0.4"])
+      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"])
+    end
+  else
+    s.add_dependency(%q<configliere>, ["~> 0.1"])
+    s.add_dependency(%q<bio>, ["~> 1.4"])
+    s.add_dependency(%q<scaffolder>, ["~> 0.4"])
+    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"])
+  end
+end
+

data/spec/scaffolder/binary_helper_spec.rb

@@ -0,0 +1,71 @@
+require File.join(File.dirname(__FILE__),'..','spec_helper')
+
+describe Scaffolder::BinaryHelper do
+
+  before(:each) do
+    @help_tool = Scaffolder::Tool::Help
+
+    @tool_class = Class.new(Scaffolder::Tool)
+    @tool_name = 'type'
+    Scaffolder::Tool.const_set(@tool_name.capitalize,@tool_class)
+
+    @args = OpenStruct.new({ :rest => %W|#{@tool_name} arg1 arg2| })
+  end
+
+  after(:each) do
+    Scaffolder::Tool.send(:remove_const,'Type')
+  end
+
+  subject do
+    object = Object.new
+    object.extend described_class
+    object
+  end
+
+  describe "select_tool method" do
+
+    it "should return corresponding tool subclass when requested by name" do
+      subject[@tool_name].should == @tool_class
+    end
+
+    it "should return the help tool when passed an unknown name" do
+      subject['unknown-tool'].should == @help_tool
+    end
+
+    it "should return the help tool when passed nil" do
+      subject[nil].should == @help_tool
+    end
+
+  end
+
+  it "should fetch the right tool class when requested" do
+    tool, args = subject.determine_tool(@args)
+    tool.should == @tool_class
+    args.rest.should == @args.rest[-2..-1]
+  end
+
+  it "should fetch the help tool class when no arguments passed" do
+    no_args = Hash.new
+    no_args.expects(:rest).returns([])
+
+    expected = no_args.clone
+    expected[:empty_args] = true
+
+    tool, args = subject.determine_tool(no_args)
+
+    tool.should == @help_tool
+    args.should == expected
+  end
+
+  it "should fetch the help tool class when an invalid argument is passed" do
+    args = Hash.new
+    args.expects(:rest).returns(['unknown-tool'])
+    updated_args = args.clone
+    updated_args[:unknown_tool] = 'unknown-tool'
+
+    tool, args = subject.determine_tool(args)
+
+    tool.should == @help_tool
+    args.should == updated_args
+  end
+end

data/spec/scaffolder/tool/help_spec.rb

@@ -0,0 +1,127 @@
+require File.expand_path(File.join(File.dirname(__FILE__),'..','..','spec_helper'))
+
+describe Scaffolder::Tool::Help do
+
+  it "should inherit from Scaffolder::Tool" do
+    described_class.superclass.should == Scaffolder::Tool
+  end
+
+  it "should return the description of the tool" do
+    desc = "Help information for scaffolder commands"
+    described_class.description.should == desc
+  end
+
+  USAGE = <<-MSG.gsub(/^ {6}/, '')
+      usage: scaffolder [--version] COMMAND scaffold-file sequence-file
+      [options]
+
+      Commands:
+    MSG
+
+  before(:each) do
+    @settings = Hash.new
+    @settings.stubs(:rest).returns([])
+  end
+
+  describe "execution with no command and the version argument" do
+
+    subject do
+      @settings[:version] = true
+      @settings[:empty_args] = true
+      described_class.new(@settings)
+    end
+
+    it "should not raise an error" do
+      lambda{ subject.execute }.should_not raise_error
+    end
+
+    it "should return the version number" do
+      version = File.read('VERSION').strip
+      subject.execute.should == "scaffolder tool version " + version
+    end
+
+  end
+
+  describe "execution with no command" do
+
+    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
+
+    it "should contain the usage information" do
+      subject.execute.should include(USAGE)
+    end
+
+    it "should contain each tool information" do
+      tool_subclasses.each do |cls|
+        subject.execute.should include(cls.description)
+      end
+    end
+
+  end
+
+  describe "execution with an invalid command arguments" do
+
+    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
+
+  end
+
+  describe "execution with the name of a scaffolder tool command" do
+
+    before(:each) do
+      @tool = Class.new(Scaffolder::Tool)
+      described_class.superclass.const_set('Fake',@tool)
+
+      @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
+
+    subject do
+      @settings.stubs(:rest).returns(['fake'])
+      described_class.new(@settings)
+    end
+
+    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
+
+    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
+
+  end
+
+  describe "execution with the name of a unknown command" do
+
+    subject do
+      @settings.stubs(:rest).returns(['fake'])
+      described_class.new(@settings)
+    end
+
+    it "should raise an error" do
+      lambda{ subject.execute }.should(raise_error(ArgumentError,
+        "Unknown command 'fake'.\nSee 'scaffolder help'."))
+    end
+
+  end
+
+end