genevalidatorapp 1.4.13 → 1.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 11532c47e62fd41b42b9c9ce29f1412fce64287a
-  data.tar.gz: df79d9d63cb7c01e876fb25966033eb366aad0a3
+  metadata.gz: d5a4eadc8917e5ba9cd4a86f33e26c70f8b6669e
+  data.tar.gz: 955917782eaef7878239553f37824895cbf92eb3
 SHA512:
-  metadata.gz: 01b1631e391d51d2c0dc127cf840d12e5bd63a647e1659f95dfde57854529f8394e8a529490f0cb0739999c902d3f7aaeec0285a2cd4fc155a77bda36aa4f542
-  data.tar.gz: 9b3b9970a7a20483efecea122baa35b0bf2e135b434f8085ad0b2b70a1403fa1ceff384b0e6917284d6286d94425733126bd5d35bb2780e74867313b3a329f0c
+  metadata.gz: b55edc8bcd0f93f1061b1aec961f73d5614c676a6f92656a2af12776a6a0daf6aac315f9daefc79773740cbf914c8a34503db7683d0895ae3188f6d8bf418ba3
+  data.tar.gz: d748fcf3753c9025ca278782ec7a9fe42286010593e97b6703d40c500e0d40710dc7c437dccc8c517c2e0eb18f7621e37bbbfd42703fbb6d0948928ca51086ba

data/README.md

@@ -1,5 +1,5 @@
 # GeneValidatorApp
-[![Build Status](https://travis-ci.org/wurmlab/GeneValidatorApp.svg?branch=master)](https://travis-ci.org/wurmlab/GeneValidatorApp)
+[![Build Status](https://travis-ci.org/wurmlab/genevalidatorapp.svg?branch=master)](https://travis-ci.org/wurmlab/genevalidatorapp)
 [![Gem Version](https://badge.fury.io/rb/GeneValidatorApp.svg)](http://badge.fury.io/rb/GeneValidatorApp)
 [![Dependency Status](https://gemnasium.com/wurmlab/GeneValidatorApp.svg)](https://gemnasium.com/wurmlab/GeneValidatorApp)
 [![Scrutinizer Code Quality](https://scrutinizer-ci.com/g/wurmlab/GeneValidatorApp/badges/quality-score.png?b=master)](https://scrutinizer-ci.com/g/wurmlab/GeneValidatorApp/?branch=master)

data/Rakefile

@@ -5,7 +5,11 @@ require 'rspec/core/rake_task'
 task default: [:build]
 desc 'Installs the ruby gem'
 task :build do
-  exec("gem build genevalidatorapp.gemspec && gem install ./genevalidatorapp-#{GeneValidatorApp::VERSION}.gem")
+  lib = File.expand_path('../lib', __FILE__)
+  $LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+  require 'genevalidatorapp/version'
+  exec('gem build genevalidatorapp.gemspec &&'\
+       " gem install ./genevalidatorapp-#{GeneValidatorApp::VERSION}.gem")
 end
 
 task test: :spec

data/bin/genevalidatorapp

@@ -1,4 +1,6 @@
 #!/usr/bin/env ruby
+require 'readline'
+require 'English'
 require 'slop'
 
 ENV['RACK_ENV'] ||= 'production'
@@ -6,9 +8,9 @@ ENV['RACK_ENV'] ||= 'production'
 # display name for tools like `ps`
 $PROGRAM_NAME = 'genevalidatorapp'
 
-opts = Slop.parse do |o|
-  o.banner = <<BNR
-
+begin
+  Slop.parse!(strict: true, help: true) do
+    banner <<BANNER
 SUMMARY:
   GeneValidator - Identify problems with predicted genes
 
@@ -24,99 +26,251 @@ Examples:
 
   # Create a config file with the other arguments
   $ genevalidatorapp -s -d ~/database_dir
-BNR
 
-  o.separator 'Compulsory Argument, unless set in a config file'
+BANNER
+    # - Add Web_dir ? or simply default to ~/.genevalidatorapp/runs/
 
-  o.string '-d', '--database_dir',
-           'Read BLAST database from this directory'
+    on 'c', 'config_file=',
+       'Use the given configuration file',
+       argument: true
 
-  o.separator ''
-  o.separator 'Optional Arguments'
+    on 'b', 'bin=',
+       'Load BLAST+ and/or BLAST binaries from this directory',
+       argument: true,
+       as: Array
 
-  o.string '-f', '--default_db',
-           'The Path to the the default database'
+    on 'd', 'database_dir=',
+       'Read BLAST database from this directory',
+       argument: true
 
-  o.string '-n', '--num_threads',
-           'Number of threads to use to run a BLAST search'
+    on 'f', 'default_database_path=',
+       'The path to the default BLAST database',
+       argument: true
 
-  o.string '-c', '--config_file',
-           'Use the given configuration file'
+    on 'n', 'num_threads=',
+       'Number of threads to use to run a BLAST search',
+       argument: true
 
-  o.string '-r', '--require',
-           'Load extension from this file'
+    on 'r', 'require=',
+       'Load extension from this file',
+       argument: true
 
-  o.string '--host',
-           'Host to run GeneValidatorApp on'
+    on 'H', 'host=',
+       'Host to run GeneValidatorApp on',
+       argument: true
 
-  o.string '-p', '--port',
-           'Port to run GeneValidatorApp on'
+    on 'p', 'port=',
+       'Port to run GeneValidatorApp on',
+       argument: true
 
-  o.string '-s', '--set',
-           'Set configuration value in the config file'
+    on 's', 'set',
+       'Set configuration value in default or given config file'
 
-  o.string '-l', '--list_dbs',
-           'List BLAST databases'
+    on 'l', 'list_databases',
+       'List BLAST databases'
 
-  o.string '-b', '--blast_bin',
-           'Load BLAST+ binaries from this directory'
+    on 'D', 'devel',
+       'Start GeneValidatorApp in development mode'
 
-  o.string '-m', '--mafft_bin',
-           'Load Mafft binaries from this directory'
+    on '-v', '--version',
+       'Print version number of GeneValidatorApp that will be loaded'
 
-  o.string '-w', '--web_dir',
-           'Path to the web directory (contains ' \
-           'supporting files utilised by the app).'
+    on '-h', '--help',
+       'Display this help message'
 
-  o.bool   '-D', '--devel',
-           'Start GeneValidatorApp in development mode'
+    clean_opts = lambda do |hash|
+      hash.delete_if { |k, v| k == :set || v.nil? }
+      hash
+    end
 
-  o.bool   '-v', '--version',
-           'Print version number of GeneValidatorApp that will be loaded'
+    run do
+      if version?
+        require 'genevalidatorapp/version'
+        puts GeneValidatorApp::VERSION
+        exit
+      end
 
-  o.on     '-h', '--help',
-           'Display this help message'
-end
+      ENV['RACK_ENV'] = 'development' if devel?
 
-if opts.help?
-  puts opts
-  exit
-end
+      # Exit gracefully on SIGINT.
+      stty = `stty -g`.chomp
+      trap('INT') do
+        puts ''
+        puts 'Aborted.'
+        system('stty', stty)
+        exit
+      end
 
-if opts.version?
-  require 'GeneValidatorApp/version'
-  puts GeneValidatorApp::VERSION
-  exit
-end
+      require 'genevalidatorapp'
 
-ENV['RACK_ENV'] = 'development' if opts.devel?
+      begin
+        GeneValidatorApp.init clean_opts[to_h]
 
-# Exit gracefully on SIGINT.
-stty = `stty -g`.chomp
-trap('INT') do
-  puts ''
-  puts 'Aborted.'
-  system('stty', stty)
-  exit
-end
+      # The aim of following error recovery scenarios is to guide user to a
+      # working GeneValidatorApp installation. We expect to land following
+      # error scenarios either when creating a new GeneValidatorApp (first
+      # time or later), or updating config values using -s CLI option.
 
-clean_opts = lambda do |hash|
-  hash.delete_if { |k, v| k == :set || k == :version || v.nil? }
-  hash
-end
+      rescue GeneValidatorApp::CONFIG_FILE_ERROR,
+             GeneValidatorApp::BLAST_DATABASE_ERROR => e
 
-require 'GeneValidatorApp'
-begin
-  GeneValidatorApp.init clean_opts[opts.to_hash]
-rescue SystemExit => e
-  puts '*** Error: GeneValidator failed to initialise properly.'
-  puts '    Please check all paramaters and try again.'
-  puts '    See https://github.com/IsmailM/GeneValidatorApp for more help'
-  exit e.status
-end
+        puts e
+        exit!
 
-puts GeneValidatorApp::Database.all? if opts.list_databases?
+      rescue GeneValidatorApp::BIN_DIR_NOT_FOUND => e
 
-GeneValidatorApp.send(:write_config_file) if opts.set?
+        puts e
 
-GeneValidatorApp.run
+        unless bin?
+          puts 'You can set the correct value by running:'
+          puts
+          puts '  genevalidatorapp -s -b <value>'
+          puts
+        end
+
+        exit!
+
+      rescue GeneValidatorApp::DATABASE_DIR_NOT_FOUND => e
+
+        puts e
+
+        unless database_dir?
+          puts 'You can set the correct value by running:'
+          puts
+          puts '  genevalidatorapp -s -d <value>'
+          puts
+        end
+
+        exit!
+
+      rescue GeneValidatorApp::NUM_THREADS_INCORRECT => e
+
+        puts e
+
+        unless num_threads?
+          puts 'You can set the correct value by running:'
+          puts
+          puts '  genevalidatorapp -s -n <value>'
+          puts
+        end
+
+        exit!
+
+      rescue GeneValidatorApp::EXTENSION_FILE_NOT_FOUND => e
+
+        puts e
+
+        unless require?
+          puts 'You can set the correct value by running:'
+          puts
+          puts '  genevalidatorapp -s -r <value>'
+          puts
+        end
+
+        exit!
+
+      rescue GeneValidatorApp::BLAST_NOT_INSTALLED,
+             GeneValidatorApp::BLAST_NOT_EXECUTABLE,
+             GeneValidatorApp::BLAST_NOT_COMPATIBLE => e
+
+        # Show original error message first.
+        puts
+        puts e
+
+        # Set a flag so that if we recovered from error resulting config can be
+        # saved. Config will be saved unless invoked with -b option.
+        fetch_option(:set).value = !bin?
+
+        # Ask user if she already has BLAST+ downloaded.
+        puts
+        puts <<MSG
+GeneValidatorApp can use NCBI BLAST+ that you may have on your system already,
+or download the correct package for itself.
+Please enter the path to NCBI BLAST+.
+
+Press Ctrl+C to quit.
+MSG
+        puts
+        response = Readline.readline('>> ').to_s.strip
+        unless response.empty?
+          unless File.basename(response) == 'bin'
+            response = File.join(response, 'bin')
+          end
+          fetch_option(:bin).value = File.join(response)
+          puts
+          redo
+        end
+
+      rescue GeneValidatorApp::DATABASE_DIR_NOT_SET => e
+
+        # Show original error message.
+        puts
+        puts e
+
+        # Set a flag so that if we recovered from error resulting config can be
+        # saved. Config will be saved unless invoked with -d option.
+        fetch_option(:set).value = !database_dir?
+
+        # Ask user for the directory containing sequences or BLAST+
+        # databases.
+        puts
+        puts <<MSG
+GeneValidatorApp needs to know where your BLAST+ databases are.
+Please enter the path to the relevant directory.
+
+Press Ctrl+C to quit.
+MSG
+
+        puts
+        response = Readline.readline('>> ').to_s.strip
+        fetch_option(:database_dir).value = response
+        redo
+
+      rescue GeneValidatorApp::NO_BLAST_DATABASE_FOUND => e
+        unless list_databases?
+
+          # Print error raised.
+          puts
+          puts e
+
+        end
+
+      rescue => e
+        # This will catch any unhandled error and some very special errors.
+        # Ideally we will never hit this block. If we do, there's a bug in
+        # GeneValidatorApp or something really weird going on. If we hit this
+        # error block we show the stacktrace to the user requesting them to
+        # post the same to our Google Group.
+        puts <<MSG
+Something went wonky
+
+Looks like you have encountered a bug in GeneValidatorApp. Please could you
+report this incident here -
+https://github.com/wurmlab/genevalidatorapp/issues
+
+Error:
+#{e.backtrace.unshift(e.message).join("\n")}
+MSG
+        exit
+      end
+
+      if list_databases?
+        puts GeneValidatorApp::Database.all
+        exit
+      end
+
+      if set?
+        GeneValidatorApp.config.write_config_file
+        exit
+      end
+
+      GeneValidatorApp.config.write_config_file if fetch_option(:set).value
+
+      GeneValidatorApp.run
+    end
+  end
+rescue Slop::Error => e
+  puts e
+  puts "Run '#{$PROGRAM_NAME} -h' for help with command line options."
+  exit
+end

data/config.ru

@@ -1,3 +1,3 @@
-require 'GeneValidatorApp'
+require 'genevalidatorapp'
 GeneValidatorApp.init
 run GeneValidatorApp

data/genevalidatorapp.gemspec

@@ -1,7 +1,7 @@
 # coding: utf-8
 lib = File.expand_path('../lib', __FILE__)
 $LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
-require 'GeneValidatorApp/version'
+require 'genevalidatorapp/version'
 
 Gem::Specification.new do |spec|
   spec.name          = 'genevalidatorapp'
@@ -9,8 +9,9 @@ Gem::Specification.new do |spec|
   spec.authors       = ['Ismail Moghul']
   spec.email         = ['Ismail.Moghul@gmail.com']
   spec.summary       = 'A Web App wrapper for GeneValidator.'
-  spec.description   = 'A Web App wrapper for GeneValidator, a program for validating gene predictions.'
-  spec.homepage      = 'https://github.com/wurmlab/GeneValidatorApp'
+  spec.description   = 'A Web App wrapper for GeneValidator, a program for' \
+                       ' validating gene predictions.'
+  spec.homepage      = 'https://github.com/wurmlab/genevalidatorapp'
   spec.license       = 'AGPL'
 
   spec.files         = `git ls-files -z`.split("\x0")
@@ -25,13 +26,13 @@ Gem::Specification.new do |spec|
   spec.add_development_dependency 'rspec', '~> 3.2'
   spec.add_development_dependency 'rack-test', '~> 0.6'
 
+  spec.add_dependency 'genevalidator', '~>1.6'
   spec.add_dependency 'bio', '~>1.4'
-  spec.add_dependency 'genevalidator', '~>1.5'
   spec.add_dependency 'sinatra', '~>1.4'
   spec.add_dependency 'sinatra-contrib', '~>1.4'
   spec.add_dependency 'sinatra-cross_origin', '~> 0.3'
   spec.add_dependency 'slim', '~>3.0'
-  spec.add_dependency 'slop', '~>4.0'
+  spec.add_dependency 'slop', '~>3.6'
   spec.add_dependency 'thin', '~>1.6'
   spec.add_dependency 'w3c_validators', '~>1.1'
   spec.post_install_message = <<INFO
@@ -43,7 +44,7 @@ Gem::Specification.new do |spec|
 
     $ genevalidatorapp [options]
 
-  Visit https://github.com/wurmlab/GeneValidatorApp for more information.
+  Visit https://github.com/wurmlab/genevalidatorapp for more information.
 ------------------------------------------------------------------------
 
 INFO

data/lib/genevalidatorapp.rb

@@ -0,0 +1,247 @@
+require 'yaml'
+require 'fileutils'
+
+require 'genevalidatorapp/config'
+require 'genevalidatorapp/database'
+require 'genevalidatorapp/exceptions'
+require 'genevalidatorapp/genevalidator'
+require 'genevalidatorapp/logger'
+require 'genevalidatorapp/routes'
+require 'genevalidatorapp/server'
+require 'genevalidatorapp/version'
+
+module GeneValidatorApp
+  # Use a fixed minimum version of BLAST+
+  MINIMUM_BLAST_VERSION = '2.2.30+'
+
+  class << self
+    def environment
+      ENV['RACK_ENV']
+    end
+
+    def verbose?
+      @verbose ||= (environment == 'development')
+    end
+
+    def root
+      File.dirname(File.dirname(__FILE__))
+    end
+
+    def logger
+      @logger ||= Logger.new(STDERR, verbose?)
+    end
+
+    # Setting up the environment before running the app...
+    def init(config = {})
+      @config = Config.new(config)
+
+      init_binaries
+      init_database
+
+      check_dirs
+      init_gv_tempdir
+      init_public_dir
+
+      load_extension
+      check_num_threads
+      check_max_characters
+      self
+    end
+
+    attr_reader :config, :temp_dir, :public_dir
+
+    # Starting the app manually
+    def run
+      check_host
+      Server.run(self)
+    rescue Errno::EADDRINUSE
+      puts "** Could not bind to port #{config[:port]}."
+      puts "   Is GeneValidator already accessible at #{server_url}?"
+      puts '   No? Try running GeneValidator on another port, like so:'
+      puts
+      puts '       genevalidatorapp -p 4570.'
+    rescue Errno::EACCES
+      puts "** Need root privilege to bind to port #{config[:port]}."
+      puts '   It is not advisable to run GeneValidator as root.'
+      puts '   Please use Apache/Nginx to bind to a privileged port.'
+    end
+
+    def on_start
+      puts '** GeneValidator is ready.'
+      puts "   Go to #{server_url} in your browser and start analysing genes!"
+      puts '   Press CTRL+C to quit.'
+      open_in_browser(server_url)
+    end
+
+    def on_stop
+      puts
+      puts '** Thank you for using GeneValidatorApp :).'
+      puts '   Please cite: '
+      puts '        Dragan M., Moghul M.I., Priyam A., Wurm Y (in prep).'
+      puts '        GeneValidator: identify problematic gene predictions.'
+    end
+
+    # Rack-interface.
+    #
+    # Inject our logger in the env and dispatch request to our controller.
+    def call(env)
+      env['rack.logger'] = logger
+      Routes.call(env)
+    end
+
+    private
+
+    def check_dirs
+      config[:gv_app_dir] = File.expand_path(config[:gv_app_dir])
+      unique_run_id       = 'GV_' + "#{Time.now.strftime('%Y%m%d-%H-%M-%S')}"
+      @public_dir         = File.join(config[:gv_app_dir], unique_run_id)
+      FileUtils.mkdir_p(File.join(@public_dir, 'GeneValidator'))
+    end
+
+    # Creates a Temp directory (starting with 'GeneValidator_') each time
+    #   GVapp is started. Within this Temp folder, sub directories are created
+    #   in which GeneValidator is run.
+    def init_gv_tempdir
+      @temp_dir = Dir.mktmpdir('GeneValidator_')
+    end
+
+    # Copy the public folder (in the app root) to the gv_app_dir location - this
+    #   gv_app_dir is then used by the app to serve all dependencies...
+    def init_public_dir
+      root_web_files_dir  = File.join(GeneValidatorApp.root, 'public/web_files')
+      FileUtils.cp_r(root_web_files_dir, @public_dir)
+    end
+
+    def init_binaries
+      config[:bin] = init_bins if config[:bin]
+      assert_blast_installed_and_compatible
+      assert_mafft_installed
+    end
+
+    def init_database
+      fail DATABASE_DIR_NOT_SET unless config[:database_dir]
+
+      config[:database_dir] = File.expand_path(config[:database_dir])
+      unless File.exist?(config[:database_dir]) &&
+             File.directory?(config[:database_dir])
+        fail DATABASE_DIR_NOT_FOUND, config[:database_dir]
+      end
+
+      assert_blast_databases_present_in_database_dir
+      logger.debug("Will use BLAST+ databases at: #{config[:database_dir]}")
+
+      Database.scan_databases_dir
+      Database.each do |database|
+        logger.debug("Found #{database.type.chomp} database" \
+                     " '#{database.title.chomp}' at '#{database.name.chomp}'")
+      end
+    end
+
+    def load_extension
+      return unless config[:require]
+      config[:require] = File.expand_path config[:require]
+      unless File.exist?(config[:require]) && File.file?(config[:require])
+        fail EXTENSION_FILE_NOT_FOUND, config[:require]
+      end
+
+      logger.debug("Loading extension: #{config[:require]}")
+      require config[:require]
+    end
+
+    def assert_blast_databases_present_in_database_dir
+      cmd = "blastdbcmd -recursive -list #{config[:database_dir]}"
+      out = `#{cmd}`
+      errpat = /BLAST Database error/
+      fail NO_BLAST_DATABASE_FOUND, config[:database_dir] if out.empty?
+      fail BLAST_DATABASE_ERROR, cmd, out if out.match(errpat) ||
+                                             !$CHILD_STATUS.success?
+    end
+
+    def check_num_threads
+      num_threads = Integer(config[:num_threads])
+      fail NUM_THREADS_INCORRECT unless num_threads > 0
+
+      logger.debug "Will use #{num_threads} threads to run BLAST."
+      if num_threads > 256
+        logger.warn "Number of threads set at #{num_threads} is unusually high."
+      end
+    rescue
+      raise NUM_THREADS_INCORRECT
+    end
+
+    def check_max_characters
+      if config[:max_characters] != 'undefined'
+        config[:max_characters] = Integer(config[:max_characters])
+      end
+    rescue
+      raise MAX_CHARACTERS_INCORRECT
+    end
+
+    def init_bins
+      bins = []
+      config[:bin].each do |bin|
+        bins << File.expand_path(bin)
+        unless File.exist?(bin) && File.directory?(bin)
+          fail BIN_DIR_NOT_FOUND, config[:bin]
+        end
+        export_bin_dir(bin)
+      end
+      bins
+    end
+
+    ## Checks if dir is in $PATH and if not, it adds the dir to the $PATH.
+    def export_bin_dir(bin_dir)
+      return unless bin_dir
+      return if ENV['PATH'].split(':').include?(bin_dir)
+      ENV['PATH'] = "#{bin_dir}:#{ENV['PATH']}"
+    end
+
+    def assert_blast_installed_and_compatible
+      fail BLAST_NOT_INSTALLED unless command? 'blastdbcmd'
+      version = `blastdbcmd -version`.split[1]
+      fail BLAST_NOT_COMPATIBLE, version unless version >= MINIMUM_BLAST_VERSION
+    end
+
+    def assert_mafft_installed
+      fail MAFFT_NOT_INSTALLED unless command? 'mafft'
+    end
+
+    # Check and warn user if host is 0.0.0.0 (default).
+    def check_host
+      # rubocop:disable Style/GuardClause
+      if config[:host] == '0.0.0.0'
+        logger.warn 'Will listen on all interfaces (0.0.0.0).' \
+                    ' Consider using 127.0.0.1 (--host option).'
+      end
+      # rubocop:enable Style/GuardClause
+    end
+
+    def server_url
+      host = config[:host]
+      host = 'localhost' if host == '127.0.0.1' || host == '0.0.0.0'
+      "http://#{host}:#{config[:port]}"
+    end
+
+    def open_in_browser(server_url)
+      return if using_ssh? || verbose?
+      if RUBY_PLATFORM =~ /linux/ && xdg?
+        `xdg-open #{server_url}`
+      elsif RUBY_PLATFORM =~ /darwin/
+        `open #{server_url}`
+      end
+    end
+
+    def using_ssh?
+      true if ENV['SSH_CLIENT'] || ENV['SSH_TTY'] || ENV['SSH_CONNECTION']
+    end
+
+    def xdg?
+      true if ENV['DISPLAY'] && command?('xdg-open')
+    end
+
+    # Return `true` if the given command exists and is executable.
+    def command?(command)
+      system("which #{command} > /dev/null 2>&1")
+    end
+  end
+end