genomer-plugin-summary 0.0.1 → 0.0.2

data/VERSION

@@ -1 +1 @@
-0.0.1
+0.0.2

data/features/sequences.feature

@@ -0,0 +1,246 @@
+Feature: Producing a summary of the scaffold sequences
+  In order to have an overview of the sequences in a scaffold
+  A user can use the "sequence" command
+  to generate the a tabular output of the scaffold sequences
+
+  @disable-bundler
+  Scenario: An empty scaffold
+    Given I create a new genomer project
+      And I write to "assembly/scaffold.yml" with:
+      """
+      ---
+        -
+          unresolved:
+            length: 50
+      """
+      And I write to "assembly/sequence.fna" with:
+      """
+      >contig0001
+      ATGC
+      """
+     When I run `genomer summary sequences`
+     Then the exit status should be 0
+      And the output should contain:
+      """
+      +------------------+------------+------------+------------+----------+--------+
+      |                             Scaffold Sequences                              |
+      +------------------+------------+------------+------------+----------+--------+
+      | Sequence         | Start (bp) |  End (bp)  | Size (bp)  | Size (%) | GC (%) |
+      +------------------+------------+------------+------------+----------+--------+
+      +------------------+------------+------------+------------+----------+--------+
+      | All              |         NA |         NA |         NA |       NA |     NA |
+      +------------------+------------+------------+------------+----------+--------+
+      """
+
+  Scenario: A scaffold with a single sequence
+    Given I create a new genomer project
+      And I write to "assembly/scaffold.yml" with:
+      """
+      ---
+        -
+          sequence:
+            source: contig0001
+      """
+      And I write to "assembly/sequence.fna" with:
+      """
+      >contig0001
+      ATGC
+      """
+     When I run `genomer summary sequences`
+     Then the exit status should be 0
+      And the output should contain:
+      """
+      +------------------+------------+------------+------------+----------+--------+
+      |                             Scaffold Sequences                              |
+      +------------------+------------+------------+------------+----------+--------+
+      | Sequence         | Start (bp) |  End (bp)  | Size (bp)  | Size (%) | GC (%) |
+      +------------------+------------+------------+------------+----------+--------+
+      | contig0001       |          1 |          4 |          4 |   100.00 |  50.00 |
+      +------------------+------------+------------+------------+----------+--------+
+      | All              |          1 |          4 |          4 |   100.00 |  50.00 |
+      +------------------+------------+------------+------------+----------+--------+
+      """
+
+  Scenario: A scaffold with a two different sequences
+    Given I create a new genomer project
+      And I write to "assembly/scaffold.yml" with:
+      """
+      ---
+        -
+          sequence:
+            source: contig0001
+        -
+          sequence:
+            source: contig0002
+      """
+      And I write to "assembly/sequence.fna" with:
+      """
+      >contig0001
+      ATGCGC
+      >contig0002
+      ATATGC
+      """
+     When I run `genomer summary sequences`
+     Then the exit status should be 0
+      And the output should contain:
+      """
+      +------------------+------------+------------+------------+----------+--------+
+      |                             Scaffold Sequences                              |
+      +------------------+------------+------------+------------+----------+--------+
+      | Sequence         | Start (bp) |  End (bp)  | Size (bp)  | Size (%) | GC (%) |
+      +------------------+------------+------------+------------+----------+--------+
+      | contig0001       |          1 |          6 |          6 |    50.00 |  66.67 |
+      | contig0002       |          7 |         12 |          6 |    50.00 |  33.33 |
+      +------------------+------------+------------+------------+----------+--------+
+      | All              |          1 |         12 |         12 |   100.00 |  50.00 |
+      +------------------+------------+------------+------------+----------+--------+
+      """
+
+  Scenario: A scaffold with a two repeated sequences
+    Given I create a new genomer project
+      And I write to "assembly/scaffold.yml" with:
+      """
+      ---
+        -
+          sequence:
+            source: contig0001
+        -
+          sequence:
+            source: contig0001
+      """
+      And I write to "assembly/sequence.fna" with:
+      """
+      >contig0001
+      ATGCGC
+      """
+     When I run `genomer summary sequences`
+     Then the exit status should be 0
+      And the output should contain:
+      """
+      +------------------+------------+------------+------------+----------+--------+
+      |                             Scaffold Sequences                              |
+      +------------------+------------+------------+------------+----------+--------+
+      | Sequence         | Start (bp) |  End (bp)  | Size (bp)  | Size (%) | GC (%) |
+      +------------------+------------+------------+------------+----------+--------+
+      | contig0001       |          1 |          6 |          6 |    50.00 |  66.67 |
+      | contig0001       |          7 |         12 |          6 |    50.00 |  66.67 |
+      +------------------+------------+------------+------------+----------+--------+
+      | All              |          1 |         12 |         12 |   100.00 |  66.67 |
+      +------------------+------------+------------+------------+----------+--------+
+      """
+
+  Scenario: A scaffold with a two sequences separated by a gap
+    Given I create a new genomer project
+      And I write to "assembly/scaffold.yml" with:
+      """
+      ---
+        -
+          sequence:
+            source: contig0001
+        -
+          unresolved:
+            length: 8
+        -
+          sequence:
+            source: contig0002
+      """
+      And I write to "assembly/sequence.fna" with:
+      """
+      >contig0001
+      ATGCGC
+      >contig0002
+      ATATGC
+      """
+     When I run `genomer summary sequences`
+     Then the exit status should be 0
+      And the output should contain:
+      """
+      +------------------+------------+------------+------------+----------+--------+
+      |                             Scaffold Sequences                              |
+      +------------------+------------+------------+------------+----------+--------+
+      | Sequence         | Start (bp) |  End (bp)  | Size (bp)  | Size (%) | GC (%) |
+      +------------------+------------+------------+------------+----------+--------+
+      | contig0001       |          1 |          6 |          6 |    30.00 |  66.67 |
+      | contig0002       |         15 |         20 |          6 |    30.00 |  33.33 |
+      +------------------+------------+------------+------------+----------+--------+
+      | All              |          1 |         20 |         12 |    60.00 |  50.00 |
+      +------------------+------------+------------+------------+----------+--------+
+      """
+
+  Scenario: A scaffold with a two sequences and a gap at the start
+    Given I create a new genomer project
+      And I write to "assembly/scaffold.yml" with:
+      """
+      ---
+        -
+          unresolved:
+            length: 8
+        -
+          sequence:
+            source: contig0001
+        -
+          sequence:
+            source: contig0002
+      """
+      And I write to "assembly/sequence.fna" with:
+      """
+      >contig0001
+      ATGCGC
+      >contig0002
+      ATATGC
+      """
+     When I run `genomer summary sequences`
+     Then the exit status should be 0
+      And the output should contain:
+      """
+      +------------------+------------+------------+------------+----------+--------+
+      |                             Scaffold Sequences                              |
+      +------------------+------------+------------+------------+----------+--------+
+      | Sequence         | Start (bp) |  End (bp)  | Size (bp)  | Size (%) | GC (%) |
+      +------------------+------------+------------+------------+----------+--------+
+      | contig0001       |          9 |         14 |          6 |    30.00 |  66.67 |
+      | contig0002       |         15 |         20 |          6 |    30.00 |  33.33 |
+      +------------------+------------+------------+------------+----------+--------+
+      | All              |          9 |         20 |         12 |    60.00 |  50.00 |
+      +------------------+------------+------------+------------+----------+--------+
+      """
+
+
+  Scenario: A scaffold with a two sequences and a gap at the end
+    Given I create a new genomer project
+      And I write to "assembly/scaffold.yml" with:
+      """
+      ---
+        -
+          sequence:
+            source: contig0001
+        -
+          sequence:
+            source: contig0002
+        -
+          unresolved:
+            length: 8
+      """
+      And I write to "assembly/sequence.fna" with:
+      """
+      >contig0001
+      ATGCGC
+      >contig0002
+      ATATGC
+      """
+     When I run `genomer summary sequences`
+     Then the exit status should be 0
+      And the output should contain:
+      """
+      +------------------+------------+------------+------------+----------+--------+
+      |                             Scaffold Sequences                              |
+      +------------------+------------+------------+------------+----------+--------+
+      | Sequence         | Start (bp) |  End (bp)  | Size (bp)  | Size (%) | GC (%) |
+      +------------------+------------+------------+------------+----------+--------+
+      | contig0001       |          1 |          6 |          6 |    30.00 |  66.67 |
+      | contig0002       |          7 |         12 |          6 |    30.00 |  33.33 |
+      +------------------+------------+------------+------------+----------+--------+
+      | All              |          1 |         12 |         12 |    60.00 |  50.00 |
+      +------------------+------------+------------+------------+----------+--------+
+      """
+

data/lib/genomer-plugin-summary/gaps.rb

@@ -39,7 +39,7 @@ class GenomerPluginSummary::Gaps < Genomer::Plugin
     table.align_column 3, :right
     table.align_column 4, :center
 
-    table
+    table.to_s
   end
 
   def gap_locations(seq)

data/lib/genomer-plugin-summary/sequences.rb

@@ -0,0 +1,95 @@
+require 'genomer'
+require 'terminal-table'
+
+class GenomerPluginSummary::Sequences < Genomer::Plugin
+
+  def run
+    sequences = calculate(scaffold)
+    total     = total(sequences)
+
+    tabulate(sequences,total)
+  end
+
+  def headings
+    ['Sequence'.left(16),
+     'Start (bp)'.center(10),
+     'End (bp)'.center(10),
+     'Size (bp)'.center(10),
+     'Size (%)'.center(8),
+     'GC (%)'.center(6)]
+  end
+
+  def title
+    'Scaffold Sequences'
+  end
+
+  def tabulate(rows,total)
+    table = Terminal::Table.new(:title => title) do |t|
+      t << headings
+      t << :separator
+      rows.each do |row|
+        t << table_array(row)
+      end
+      t << :separator
+      t << table_array(total.merge({:sequence => 'All'}))
+    end
+
+    table.align_column 0, :left
+    table.align_column 1, :right
+    table.align_column 2, :right
+    table.align_column 3, :right
+    table.align_column 4, :right
+    table.align_column 5, :right
+
+    table.to_s
+  end
+
+  def table_array(hash)
+    [:sequence,:start,:end,:size,:percent,:gc].
+      map{|i| hash[i]}.
+      map{|i| i.class == Float ? sprintf('%#.2f',i) : i }
+  end
+
+  def calculate(scaffold)
+    total_length = scaffold.map(&:sequence).join.length.to_f
+
+    length = 0
+    scaffold.map do |entry|
+      i = nil
+      if entry.entry_type != :unresolved
+        entry_length = entry.sequence.length
+        i = { :sequence => entry.source,
+              :start    => length + 1, 
+              :end      => length + entry_length,
+              :size     => entry_length,
+              :percent  => entry_length / total_length * 100,
+              :gc       => gc_content(entry.sequence) }
+      end
+        
+      length += entry.sequence.length
+      i
+    end.compact
+  end
+
+  def total(seqs)
+    return Hash[[:start, :end, :size, :percent, :gc].map{|i| [i, 'NA']}] if seqs.empty?
+
+    totals = seqs.inject({:size => 0, :percent => 0, :gc => 0}) do |hash,entry|
+      hash[:start]  ||= entry[:start]
+      hash[:end]      = entry[:end]
+      hash[:size]    += entry[:size]
+      hash[:percent] += entry[:percent]
+      hash[:gc]      += entry[:gc] * entry[:size]
+
+      hash
+    end
+    totals[:gc] /= totals[:size]
+    totals
+  end
+
+  def gc_content(sequence)
+    nucleotides = sequence.gsub(/[^ATGCatgc]/,'')
+    nucleotides.gsub(/[^GCgc]/,'').length.to_f / nucleotides.length * 100
+  end
+
+end

data/spec/genomer-plugin-summary_spec/gaps_spec.rb

@@ -6,7 +6,7 @@ describe GenomerPluginSummary::Gaps do
   describe "#tabulate" do
 
     subject do
-      described_class.new([],{}).tabulate(contigs).to_s + "\n"
+      described_class.new([],{}).tabulate(contigs) + "\n"
     end
 
     context "passed an empty array" do
@@ -74,20 +74,6 @@ describe GenomerPluginSummary::Gaps do
 
   describe "#determine_gaps" do
 
-    def sequence(seq)
-      s = mock!
-      stub(s).sequence{ seq }
-      stub(s).entry_type{ :sequence }
-      s
-    end
-
-    def unresolved(seq)
-      s = mock!
-      stub(s).sequence{ seq }
-      stub(s).entry_type{ :unresolved }
-      s
-    end
-
     subject do
       described_class.new([],{}).determine_gaps(scaffold)
     end

data/spec/genomer-plugin-summary_spec/sequences_spec.rb

@@ -0,0 +1,235 @@
+require 'spec_helper'
+require 'genomer-plugin-summary/sequences'
+
+describe GenomerPluginSummary::Sequences do
+
+  def row(name,start,stop,percent,gc)
+    {:sequence => name,
+     :start    => start,
+     :end      => stop,
+     :size     => (stop - start) + 1,
+     :percent  => percent,
+     :gc       => gc}
+  end
+
+  describe "#tabulate" do
+
+    subject do
+      described_class.new([],{}).tabulate(sequences,total) + "\n"
+    end
+
+    context "passed an empty array" do
+
+      let(:sequences) do
+        []
+      end
+
+      let(:total) do
+        {:start   => 'NA',
+         :end     => 'NA',
+         :size    => 'NA',
+         :percent => 'NA',
+         :gc      => 'NA' }
+      end
+
+      it do
+        should ==<<-EOS.unindent!
+      +------------------+------------+------------+------------+----------+--------+
+      |                             Scaffold Sequences                              |
+      +------------------+------------+------------+------------+----------+--------+
+      | Sequence         | Start (bp) |  End (bp)  | Size (bp)  | Size (%) | GC (%) |
+      +------------------+------------+------------+------------+----------+--------+
+      +------------------+------------+------------+------------+----------+--------+
+      | All              |         NA |         NA |         NA |       NA |     NA |
+      +------------------+------------+------------+------------+----------+--------+
+        EOS
+      end
+
+    end
+
+    context "passed an array with a single row" do
+
+      let(:sequences) do
+        [{:sequence   => 'contig1',
+          :start      => '1',
+          :end        => '4',
+          :size       => '4',
+          :percent    => 100.0,
+          :gc         => 50.0 }]
+      end
+
+      let(:total) do
+        {:start   => '1',
+         :end     => '4',
+         :size    => '4',
+         :percent => 100.0,
+         :gc      => 50.0 }
+      end
+
+      it do
+        should ==<<-EOS.unindent!
+      +------------------+------------+------------+------------+----------+--------+
+      |                             Scaffold Sequences                              |
+      +------------------+------------+------------+------------+----------+--------+
+      | Sequence         | Start (bp) |  End (bp)  | Size (bp)  | Size (%) | GC (%) |
+      +------------------+------------+------------+------------+----------+--------+
+      | contig1          |          1 |          4 |          4 |   100.00 |  50.00 |
+      +------------------+------------+------------+------------+----------+--------+
+      | All              |          1 |          4 |          4 |   100.00 |  50.00 |
+      +------------------+------------+------------+------------+----------+--------+
+        EOS
+      end
+
+    end
+
+    context "passed a array with two rows" do
+
+      let(:sequences) do
+        [{:sequence   => 'contig1',
+          :start      => '1',
+          :end        => '4',
+          :size       => '4',
+          :percent    => 100.0,
+          :gc         => 50.0 },
+         {:sequence   => 'contig2',
+          :start      => '1',
+          :end        => '4',
+          :size       => '4',
+          :percent    => 100.0,
+          :gc         => 50.0 }]
+      end
+
+      let(:total) do
+        {:start   => '1',
+         :end     => '4',
+         :size    => '4',
+         :percent => 100.0,
+         :gc      => 50.0 }
+      end
+
+      it do
+        should ==<<-EOS.unindent!
+      +------------------+------------+------------+------------+----------+--------+
+      |                             Scaffold Sequences                              |
+      +------------------+------------+------------+------------+----------+--------+
+      | Sequence         | Start (bp) |  End (bp)  | Size (bp)  | Size (%) | GC (%) |
+      +------------------+------------+------------+------------+----------+--------+
+      | contig1          |          1 |          4 |          4 |   100.00 |  50.00 |
+      | contig2          |          1 |          4 |          4 |   100.00 |  50.00 |
+      +------------------+------------+------------+------------+----------+--------+
+      | All              |          1 |          4 |          4 |   100.00 |  50.00 |
+      +------------------+------------+------------+------------+----------+--------+
+        EOS
+      end
+
+    end
+
+  end
+
+  describe "#calculate" do
+
+    subject do
+      described_class.new([],{}).calculate(scaffold)
+    end
+
+    context "passed an empty array" do
+      let(:scaffold) do
+        []
+      end
+
+      it do
+        should == []
+      end
+    end
+
+    context "passed one sequence" do
+      let(:scaffold) do
+        [sequence('AAAGGG','contig1')]
+      end
+
+      it do
+        should == [row('contig1',1,6,100.0,50.0)]
+      end
+    end
+
+    context "passed two sequences" do
+      let(:scaffold) do
+        [sequence('AAAGGG','contig1'),
+         sequence('AAAGGG','contig2')]
+      end
+
+      it do
+        should == [row('contig1', 1,  6, 50.0, 50.0),
+                   row('contig2', 7, 12, 50.0, 50.0)]
+      end
+    end
+
+    context "passed two sequences separated by a gap" do
+      let(:scaffold) do
+        [sequence('AAAGGG','contig1'),
+         unresolved('NNNNNNNN'),
+         sequence('AAAGGG','contig2')]
+      end
+
+      it do
+        should == [row('contig1', 1,  6,  30.0, 50.0),
+                   row('contig2', 15, 20, 30.0, 50.0)]
+      end
+    end
+
+  end
+
+  describe "#total" do
+
+    subject do
+      described_class.new([],{}).total(sequences)
+    end
+
+    context "passed an empty array" do
+      let(:sequences) do
+        []
+      end
+
+      it do
+        should == {
+          :start   => 'NA',
+          :end     => 'NA',
+          :size    => 'NA',
+          :percent => 'NA',
+          :gc      => 'NA' }
+      end
+    end
+
+    context "passed one entry" do
+      let(:sequences) do
+        [row('contig1',1,6,100.0,50.0)]
+      end
+
+      it do
+        should == {
+          :start   => 1,
+          :end     => 6,
+          :size    => 6,
+          :percent => 100.0,
+          :gc      => 50.0 }
+      end
+    end
+
+    context "passed two entries less than 100% of the scaffold" do
+      let(:sequences) do
+        [row('contig1', 1,  6,  30.0, 50.0),
+         row('contig2', 15, 20, 30.0, 50.0)]
+      end
+
+      it do
+        should == {
+          :start   => 1,
+          :end     => 20,
+          :size    => 12,
+          :percent => 60.0,
+          :gc      => 50.0 }
+      end
+    end
+  end
+
+end

data/spec/spec_helper.rb

@@ -12,4 +12,20 @@ end
 
 RSpec.configure do |config|
   config.mock_with :rr
+
+  def sequence(seq, source=nil)
+    s = mock!
+    stub(s).sequence{ seq }
+    stub(s).source{ source } if source
+    stub(s).entry_type{ :sequence }
+    s
+  end
+
+  def unresolved(seq)
+    s = mock!
+    stub(s).sequence{ seq }
+    stub(s).entry_type{ :unresolved }
+    s
+  end
+
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: genomer-plugin-summary
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 0.0.2
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-05-28 00:00:00.000000000 Z
+date: 2012-10-12 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: genomer
@@ -153,15 +153,18 @@ files:
 - Rakefile
 - VERSION
 - features/gaps.feature
+- features/sequences.feature
 - features/support/env.rb
 - features/support/genomer_steps.rb
 - genomer-plugin-summary.gemspec
 - lib/genomer-plugin-summary.rb
 - lib/genomer-plugin-summary/gaps.rb
+- lib/genomer-plugin-summary/sequences.rb
 - man/genomer-summary-gaps.ronn
 - man/genomer-summary.ronn
 - spec/genomer-plugin-summary_spec.rb
 - spec/genomer-plugin-summary_spec/gaps_spec.rb
+- spec/genomer-plugin-summary_spec/sequences_spec.rb
 - spec/spec_helper.rb
 homepage: ''
 licenses: []
@@ -177,7 +180,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
       version: '0'
       segments:
       - 0
-      hash: -2942156995405587810
+      hash: 1551182150525719363
 required_rubygems_version: !ruby/object:Gem::Requirement
   none: false
   requirements:
@@ -186,7 +189,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
       segments:
       - 0
-      hash: -2942156995405587810
+      hash: 1551182150525719363
 requirements: []
 rubyforge_project: 
 rubygems_version: 1.8.23
@@ -195,8 +198,10 @@ specification_version: 3
 summary: Generates reports on the status of the genomer project
 test_files:
 - features/gaps.feature
+- features/sequences.feature
 - features/support/env.rb
 - features/support/genomer_steps.rb
 - spec/genomer-plugin-summary_spec.rb
 - spec/genomer-plugin-summary_spec/gaps_spec.rb
+- spec/genomer-plugin-summary_spec/sequences_spec.rb
 - spec/spec_helper.rb