bioinform 0.1.17 → 0.2.0

data/spec/spec_helper.rb

@@ -6,8 +6,8 @@ require 'rspec-given'
 
 require 'fileutils'
 require 'stringio'
-require 'fabrication'
+# require 'fabrication'
 
 require 'fakefs/spec_helpers'
 require 'spec_helper_source'
-Fabrication.clear_definitions
+# Fabrication.clear_definitions

data/spec/spec_helper_source.rb

@@ -27,18 +27,17 @@ def capture_stderr(&block)
   capture_io(&block)[:stderr]
 end
 
-def parser_specs(parser_klass, good_cases, bad_cases)
+def parser_specs(parser, good_cases, bad_cases)
   context '#parse!' do
     good_cases.each do |case_description, input_and_result|
-      it "should be able to parse #{case_description}" do
-        result = parser_klass.new(input_and_result[:input]).parse
-        Bioinform::PM.new(result).should == input_and_result[:result]
+      it "should be able to parse #{case_description} (input #{input_and_result[:input].inspect} --> result #{input_and_result[:result].inspect})" do
+        expect(parser.parse(input_and_result[:input])).to eq input_and_result[:result]
       end
     end
 
     bad_cases.each do |case_description, input|
       it "should raise an exception on parsing #{case_description}" do
-        expect{ parser_klass.new(input[:input]).parse! }.to raise_error
+        expect{ parser.parse!(input[:input]) }.to raise_error
       end
     end
   end
@@ -57,3 +56,24 @@ end
 def make_model_file(motif, model_type)
   make_file(motif_filename(motif, model_type), motif.send(model_type))
 end
+
+RSpec::Matchers.define :be_within_range_from_matrix do |expected, range|
+  match do |actual|
+    expected.alphabet == actual.alphabet && matrices_within_range?(actual.matrix, expected.matrix, range)
+  end
+
+  def matrices_within_range?(actual, expected, range)
+    (actual.length == expected.length) && actual.each_index.all?{|pos_index| postions_within_range?(actual[pos_index], expected[pos_index], range) }
+  end
+
+  def postions_within_range?(actual, expected, range)
+    (actual.length == expected.length) && actual.each_index.all?{|letter_index| (actual[letter_index] - expected[letter_index]).abs <= range }
+  end
+
+  description do
+    "\nbe in range #{range} from\n#{expected}\n"
+  end
+  failure_message do |actual|
+    "expected that\n#{actual}\nwould be in range #{range} from\n#{expected}\n"
+  end
+end

data/spec/support_spec.rb

@@ -0,0 +1,31 @@
+require 'bioinform/support'
+
+describe Bioinform::Support do
+  specify do
+    expect(Bioinform::Support.element_indices([:A,:C,:G,:T])).to eq( {:A=>0, :C=>1, :G=>2, :T=>3} )
+  end
+
+  specify do
+    expect(Bioinform::Support.hash_keys_permuted([0,1], :A)).to eq( {[0,1] => :A, [1,0] => :A} )
+  end
+
+  specify do
+    expect(Bioinform::Support.with_key_permutations({[0,1] => :A, [0,2] => :T})).to eq( {[0,1] => :A, [1,0] => :A, [0,2] => :T, [2,0]=>:T} )
+  end
+
+  specify do
+    expect(Bioinform::Support.various_key_cases({'a' => 2, 'C' => 3, :g =>5, :T => 8})).to eq( {'a' => 2, 'A' => 2, 'c' => 3, 'C' => 3, :g =>5, :G => 5, :T => 8, :t=>8} )
+  end
+
+  specify do
+    expect(Bioinform::Support.various_key_types({'a' => 2, 'C' => 3, :g =>5, :T => 8})).to eq( {'a' => 2, :a => 2, 'C' => 3, :C => 3, :g =>5, 'g' => 5, :T => 8, 'T'=>8} )
+  end
+
+  specify do
+    expect(Bioinform::Support.various_key_value_cases({:A => :T})).to eq( {:A => :T, :a => :t} )
+  end
+
+  specify do
+    expect(Bioinform::Support.various_key_value_types({:A => :T})).to eq( {:A => :T, 'A' => 'T'} )
+  end
+end

metadata

@@ -1,34 +1,20 @@
 --- !ruby/object:Gem::Specification
 name: bioinform
 version: !ruby/object:Gem::Version
-  version: 0.1.17
+  version: 0.2.0
 platform: ruby
 authors:
 - Ilya Vorontsov
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-05-23 00:00:00.000000000 Z
-dependencies:
-- !ruby/object:Gem::Dependency
-  name: docopt
-  requirement: !ruby/object:Gem::Requirement
-    requirements:
-    - - '='
-      - !ruby/object:Gem::Version
-        version: 0.5.0
-  type: :runtime
-  prerelease: false
-  version_requirements: !ruby/object:Gem::Requirement
-    requirements:
-    - - '='
-      - !ruby/object:Gem::Version
-        version: 0.5.0
+date: 2014-07-14 00:00:00.000000000 Z
+dependencies: []
 description: A bunch of useful classes for bioinformatics
 email:
 - prijutme4ty@gmail.com
 executables:
-- merge_into_collection
+- convert_motif
 - pcm2pwm
 - split_motifs
 extensions: []
@@ -41,80 +27,46 @@ files:
 - README.md
 - Rakefile
 - TODO.txt
-- bin/merge_into_collection
+- bin/convert_motif
 - bin/pcm2pwm
 - bin/split_motifs
 - bioinform.gemspec
 - lib/bioinform.rb
+- lib/bioinform/alphabet.rb
+- lib/bioinform/background.rb
 - lib/bioinform/cli.rb
 - lib/bioinform/cli/convert_motif.rb
-- lib/bioinform/cli/merge_into_collection.rb
 - lib/bioinform/cli/pcm2pwm.rb
 - lib/bioinform/cli/split_motifs.rb
+- lib/bioinform/conversion_algorithms.rb
 - lib/bioinform/conversion_algorithms/pcm2ppm_converter.rb
 - lib/bioinform/conversion_algorithms/pcm2pwm_converter.rb
 - lib/bioinform/conversion_algorithms/pcm2pwm_mara_converter.rb
 - lib/bioinform/conversion_algorithms/ppm2pcm_converter.rb
-- lib/bioinform/conversion_algorithms/ppm2pwm_converter.rb
+- lib/bioinform/conversion_algorithms/pwm2iupac_pwm_converter.rb
+- lib/bioinform/conversion_algorithms/pwm2pcm_converter.rb
 - lib/bioinform/data_models.rb
-- lib/bioinform/data_models/collection.rb
-- lib/bioinform/data_models/motif.rb
+- lib/bioinform/data_models/named_model.rb
 - lib/bioinform/data_models/pcm.rb
 - lib/bioinform/data_models/pm.rb
 - lib/bioinform/data_models/ppm.rb
 - lib/bioinform/data_models/pwm.rb
+- lib/bioinform/errors.rb
 - lib/bioinform/formatters.rb
-- lib/bioinform/formatters/raw_formatter.rb
+- lib/bioinform/formatters/consensus_formatter.rb
+- lib/bioinform/formatters/motif_formatter.rb
+- lib/bioinform/formatters/pretty_matrix_formatter.rb
 - lib/bioinform/formatters/transfac_formatter.rb
 - lib/bioinform/parsers.rb
-- lib/bioinform/parsers/jaspar_parser.rb
 - lib/bioinform/parsers/matrix_parser.rb
-- lib/bioinform/parsers/parser.rb
-- lib/bioinform/parsers/splittable_parser.rb
-- lib/bioinform/parsers/string_fantom_parser.rb
-- lib/bioinform/parsers/string_parser.rb
-- lib/bioinform/parsers/trivial_parser.rb
-- lib/bioinform/parsers/yaml_parser.rb
+- lib/bioinform/parsers/motif_splitter.rb
 - lib/bioinform/support.rb
-- lib/bioinform/support/advanced_scan.rb
-- lib/bioinform/support/array_product.rb
-- lib/bioinform/support/array_zip.rb
-- lib/bioinform/support/collect_hash.rb
-- lib/bioinform/support/deep_dup.rb
-- lib/bioinform/support/delete_many.rb
-- lib/bioinform/support/inverf.rb
-- lib/bioinform/support/multiline_squish.rb
-- lib/bioinform/support/parameters.rb
-- lib/bioinform/support/partial_sums.rb
-- lib/bioinform/support/same_by.rb
 - lib/bioinform/support/strip_doc.rb
-- lib/bioinform/support/third_part/active_support/core_ext/array/extract_options.rb
-- lib/bioinform/support/third_part/active_support/core_ext/hash/indifferent_access.rb
-- lib/bioinform/support/third_part/active_support/core_ext/hash/keys.rb
-- lib/bioinform/support/third_part/active_support/core_ext/module/attribute_accessors.rb
-- lib/bioinform/support/third_part/active_support/core_ext/object/try.rb
-- lib/bioinform/support/third_part/active_support/core_ext/string/access.rb
-- lib/bioinform/support/third_part/active_support/core_ext/string/behavior.rb
-- lib/bioinform/support/third_part/active_support/core_ext/string/filters.rb
-- lib/bioinform/support/third_part/active_support/core_ext/string/multibyte.rb
-- lib/bioinform/support/third_part/active_support/hash_with_indifferent_access.rb
-- lib/bioinform/support/third_part/active_support/multibyte.rb
-- lib/bioinform/support/third_part/active_support/multibyte/chars.rb
-- lib/bioinform/support/third_part/active_support/multibyte/exceptions.rb
-- lib/bioinform/support/third_part/active_support/multibyte/unicode.rb
-- lib/bioinform/support/third_part/active_support/multibyte/utils.rb
 - lib/bioinform/version.rb
+- spec/alphabet_spec.rb
+- spec/background_spec.rb
 - spec/cli/cli_spec.rb
 - spec/cli/convert_motif_spec.rb
-- spec/cli/data/merge_into_collection/GABPA_f1.pwm
-- spec/cli/data/merge_into_collection/KLF4_f2.pwm
-- spec/cli/data/merge_into_collection/SP1_f1.pwm
-- spec/cli/data/merge_into_collection/collection.txt.result
-- spec/cli/data/merge_into_collection/collection.yaml.result
-- spec/cli/data/merge_into_collection/collection_pwm.yaml.result
-- spec/cli/data/merge_into_collection/pwm_folder/GABPA_f1.pwm
-- spec/cli/data/merge_into_collection/pwm_folder/KLF4_f2.pwm
-- spec/cli/data/merge_into_collection/pwm_folder/SP1_f1.pwm
 - spec/cli/data/pcm2pwm/KLF4 f2 spaced name.pcm
 - spec/cli/data/pcm2pwm/KLF4_f2.pcm
 - spec/cli/data/pcm2pwm/KLF4_f2.pwm.result
@@ -123,9 +75,7 @@ files:
 - spec/cli/data/split_motifs/GABPA_f1.mat.result
 - spec/cli/data/split_motifs/KLF4_f2.mat.result
 - spec/cli/data/split_motifs/SP1_f1.mat.result
-- spec/cli/data/split_motifs/collection.yaml
 - spec/cli/data/split_motifs/plain_collection.txt
-- spec/cli/merge_into_collection_spec.rb
 - spec/cli/pcm2pwm_spec.rb
 - spec/cli/shared_examples/convert_motif/motif_list_empty.rb
 - spec/cli/shared_examples/convert_motif/several_motifs_specified.rb
@@ -133,35 +83,24 @@ files:
 - spec/cli/shared_examples/convert_motif/yield_help_string.rb
 - spec/cli/shared_examples/convert_motif/yield_motif_conversion_error.rb
 - spec/cli/split_motifs_spec.rb
-- spec/data_models/collection_spec.rb
-- spec/data_models/motif_spec.rb
+- spec/converters/pcm2ppm_converter_spec.rb
+- spec/converters/pcm2pwm_converter_spec.rb
+- spec/converters/ppm2pcm_converter_spec.rb
+- spec/converters/pwm2iupac_pwm_converter_spec.rb
+- spec/converters/pwm2pcm_converter_spec.rb
+- spec/data_models/named_model_spec.rb
 - spec/data_models/pcm_spec.rb
 - spec/data_models/pm_spec.rb
 - spec/data_models/ppm_spec.rb
 - spec/data_models/pwm_spec.rb
-- spec/fabricators/collection_fabricator.rb
-- spec/fabricators/motif_fabricator.rb
 - spec/fabricators/motif_formats_fabricator.rb
-- spec/fabricators/pcm_fabricator.rb
-- spec/fabricators/pm_fabricator.rb
-- spec/fabricators/ppm_fabricator.rb
-- spec/fabricators/pwm_fabricator.rb
-- spec/parsers/parser_spec.rb
-- spec/parsers/string_fantom_parser_spec.rb
-- spec/parsers/string_parser_spec.rb
-- spec/parsers/trivial_parser_spec.rb
-- spec/parsers/yaml_parser_spec.rb
+- spec/formatters/consensus_formatter_spec.rb
+- spec/formatters/raw_formatter_spec.rb
+- spec/parsers/matrix_parser_spec.rb
+- spec/parsers/motif_splitter_spec.rb
 - spec/spec_helper.rb
 - spec/spec_helper_source.rb
-- spec/support/advanced_scan_spec.rb
-- spec/support/array_product_spec.rb
-- spec/support/array_zip_spec.rb
-- spec/support/collect_hash_spec.rb
-- spec/support/delete_many_spec.rb
-- spec/support/inverf_spec.rb
-- spec/support/multiline_squish_spec.rb
-- spec/support/partial_sums_spec.rb
-- spec/support/same_by_spec.rb
+- spec/support_spec.rb
 homepage: ''
 licenses: []
 metadata: {}
@@ -181,7 +120,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.2.2
+rubygems_version: 2.3.0
 signing_key: 
 specification_version: 4
 summary: Classes for work with different input formats of positional matrices and
@@ -189,17 +128,10 @@ summary: Classes for work with different input formats of positional matrices an
   several useful extensions for Enumerable module like parametric map and callable
   symbols
 test_files:
+- spec/alphabet_spec.rb
+- spec/background_spec.rb
 - spec/cli/cli_spec.rb
 - spec/cli/convert_motif_spec.rb
-- spec/cli/data/merge_into_collection/GABPA_f1.pwm
-- spec/cli/data/merge_into_collection/KLF4_f2.pwm
-- spec/cli/data/merge_into_collection/SP1_f1.pwm
-- spec/cli/data/merge_into_collection/collection.txt.result
-- spec/cli/data/merge_into_collection/collection.yaml.result
-- spec/cli/data/merge_into_collection/collection_pwm.yaml.result
-- spec/cli/data/merge_into_collection/pwm_folder/GABPA_f1.pwm
-- spec/cli/data/merge_into_collection/pwm_folder/KLF4_f2.pwm
-- spec/cli/data/merge_into_collection/pwm_folder/SP1_f1.pwm
 - spec/cli/data/pcm2pwm/KLF4 f2 spaced name.pcm
 - spec/cli/data/pcm2pwm/KLF4_f2.pcm
 - spec/cli/data/pcm2pwm/KLF4_f2.pwm.result
@@ -208,9 +140,7 @@ test_files:
 - spec/cli/data/split_motifs/GABPA_f1.mat.result
 - spec/cli/data/split_motifs/KLF4_f2.mat.result
 - spec/cli/data/split_motifs/SP1_f1.mat.result
-- spec/cli/data/split_motifs/collection.yaml
 - spec/cli/data/split_motifs/plain_collection.txt
-- spec/cli/merge_into_collection_spec.rb
 - spec/cli/pcm2pwm_spec.rb
 - spec/cli/shared_examples/convert_motif/motif_list_empty.rb
 - spec/cli/shared_examples/convert_motif/several_motifs_specified.rb
@@ -218,32 +148,21 @@ test_files:
 - spec/cli/shared_examples/convert_motif/yield_help_string.rb
 - spec/cli/shared_examples/convert_motif/yield_motif_conversion_error.rb
 - spec/cli/split_motifs_spec.rb
-- spec/data_models/collection_spec.rb
-- spec/data_models/motif_spec.rb
+- spec/converters/pcm2ppm_converter_spec.rb
+- spec/converters/pcm2pwm_converter_spec.rb
+- spec/converters/ppm2pcm_converter_spec.rb
+- spec/converters/pwm2iupac_pwm_converter_spec.rb
+- spec/converters/pwm2pcm_converter_spec.rb
+- spec/data_models/named_model_spec.rb
 - spec/data_models/pcm_spec.rb
 - spec/data_models/pm_spec.rb
 - spec/data_models/ppm_spec.rb
 - spec/data_models/pwm_spec.rb
-- spec/fabricators/collection_fabricator.rb
-- spec/fabricators/motif_fabricator.rb
 - spec/fabricators/motif_formats_fabricator.rb
-- spec/fabricators/pcm_fabricator.rb
-- spec/fabricators/pm_fabricator.rb
-- spec/fabricators/ppm_fabricator.rb
-- spec/fabricators/pwm_fabricator.rb
-- spec/parsers/parser_spec.rb
-- spec/parsers/string_fantom_parser_spec.rb
-- spec/parsers/string_parser_spec.rb
-- spec/parsers/trivial_parser_spec.rb
-- spec/parsers/yaml_parser_spec.rb
+- spec/formatters/consensus_formatter_spec.rb
+- spec/formatters/raw_formatter_spec.rb
+- spec/parsers/matrix_parser_spec.rb
+- spec/parsers/motif_splitter_spec.rb
 - spec/spec_helper.rb
 - spec/spec_helper_source.rb
-- spec/support/advanced_scan_spec.rb
-- spec/support/array_product_spec.rb
-- spec/support/array_zip_spec.rb
-- spec/support/collect_hash_spec.rb
-- spec/support/delete_many_spec.rb
-- spec/support/inverf_spec.rb
-- spec/support/multiline_squish_spec.rb
-- spec/support/partial_sums_spec.rb
-- spec/support/same_by_spec.rb
+- spec/support_spec.rb

data/bin/merge_into_collection

@@ -1,4 +0,0 @@
-#!/usr/bin/env ruby
-
-require_relative '../lib/bioinform/cli/merge_into_collection'
-Bioinform::CLI::MergeIntoCollection.main(ARGV)

data/lib/bioinform/cli/merge_into_collection.rb

@@ -1,80 +0,0 @@
-require_relative '../../bioinform'
-require 'docopt'
-require 'shellwords'
-require 'yaml'
-
-module Bioinform
-  module CLI
-    module MergeIntoCollection
-      extend Bioinform::CLI::Helpers
-      def self.main(argv)
-        doc = <<-DOCOPT
-          Tool for merging multiple motifs into a single collection file.
-          It takes motif files or (entire collections) and creates a collection consisting of them all. By default motifs are treated simply as matrices(PM), but you can (possibly should) specify data model. Output file by default are in YAML-format but it's possible to create plain text file. YAML collections are useful if you want to provide additional information for motifs in collection with another tool, plain text is more readable by humans.
-
-          Usage:
-            merge_into_collection [options] [<pm-files>...]
-
-          Options:
-            -h --help                 Show this screen.
-            -n --name NAME            Specify name for a collection. Default filename is based on this parameter
-            -o --output-file FILE     Output file for resulting collection
-            -m --data-model MODEL     Data model: PM, PCM, PPM or PWM [default: PM]
-            -p --plain-text           Output collection of motifs in plain text (motifs separated with newlines, no additional information included).
-        DOCOPT
-
-        doc.gsub!(/^#{doc[/\A +/]}/,'')
-        options = Docopt::docopt(doc, argv: argv)
-
-        plain_text = options['--plain-text']
-        name = options['--name']
-        if options['--plain-text']
-          output_file = options['--output-file'] || set_extension(name || 'collection', 'txt')
-        else
-          output_file = options['--output-file'] || set_extension(name || 'collection', 'yaml')
-        end
-        data_model = Bioinform.const_get(options['--data-model'].upcase)
-
-        if options['<pm-files>'].empty?
-          filelist = $stdin.read.shellsplit
-        else
-          filelist = options['<pm-files>']
-        end
-        
-        filelist = filelist.map do |data_source|
-          if File.directory? data_source
-            Dir.glob(File.join(data_source, '*'))
-          elsif File.file? data_source
-            data_source
-          else
-            raise "File or directory #{data_source} can't be found"
-          end
-        end.flatten
-
-        collection = Collection.new
-        collection.name = name  if name
-
-        filelist.each do |filename|
-          data_model.split_on_motifs(File.read(filename)).each do |pm|
-            pm.name ||= File.basename(filename, File.extname(filename))
-            collection << pm
-          end
-        end
-
-        if plain_text
-          File.open(output_file, 'w') do |f|
-            collection.each(options['--data-model'].downcase) do |pm|
-              f.puts(pm.to_s + "\n\n")
-            end
-          end
-        else
-          File.open(output_file, 'w'){|f| YAML.dump(collection, f) }
-        end
-        
-      rescue Docopt::Exit => e
-        puts e.message
-      end
-
-    end
-  end
-end