miga-base 0.7.21.0 → 0.7.24.0

data/lib/miga/common/net.rb

@@ -0,0 +1,74 @@
+# frozen_string_literal: true
+
+require 'net/ftp'
+require 'open-uri'
+require 'fileutils'
+
+Net::FTP.send(:remove_const, 'FTP_PORT') # just to avoid warnings
+Net::FTP.const_set('FTP_PORT', 21)
+
+##
+# General web-access functions shared throughout MiGA.
+module MiGA::Common::Net
+  ##
+  # Returns the URL of the host +name+ (Symbol)
+  def known_hosts(name)
+    case name.to_sym
+    when :miga_online_ftp
+      'ftp://microbial-genomes.org//' # <- // to simplify chdir in connection
+    when :miga_db
+      'ftp://microbial-genomes.org/db'
+    when :miga_dist
+      'ftp://microbial-genomes.org/dist'
+    else
+      raise "Unrecognized server name: #{host}"
+    end
+  end
+
+  ##
+  # Connect to an FTP +host+ (String) or a known host name (Symbol, see
+  # +.known_hosts+)
+  def remote_connection(host)
+    host = known_hosts(host) if host.is_a?(Symbol)
+    uri = URI.parse(host)
+    raise 'Only FTP hosts are currently supported' unless uri.scheme == 'ftp'
+
+    ftp = Net::FTP.new(uri.host)
+    ftp.passive = true
+    ftp.login
+    ftp.chdir(uri.path)
+    ftp
+  end
+
+  ##
+  # Download a file via FTP using the +connection+ (returned by
+  # +.remote_connection+) with remote name +file+ into local +target+.
+  #
+  # Alternatively, +connection+ can simply be the host (String) or a recognized
+  # Symbol (see +.remote_connection+), in which case the function opens the
+  # connection automatically
+  #
+  # Reports progress to the function block with two arguments: the
+  # currently transferred size and the total file size
+  def download_file_ftp(connection, file, target)
+    # Open connection unless passed
+    close_conn = false
+    if connection.is_a?(String) || connection.is_a?(Symbol)
+      connection = remote_connection(connection)
+      close_conn = true
+    end
+
+    # Prepare download
+    FileUtils.mkdir_p(File.dirname(target))
+    filesize = connection.size(file)
+    transferred = 0
+
+    # Get in chunks of 1KiB
+    connection.getbinaryfile(file, target, 1024) do |data|
+      yield(transferred += data.size, filesize) if block_given?
+    end
+
+    # Close connection if automatically opened
+    connection.close if close_conn
+  end
+end

data/lib/miga/common/with_option.rb

@@ -0,0 +1,83 @@
+# frozen_string_literal: true
+
+##
+# Helper module including specific functions to handle objects that
+# have configurable options. The class including this module must implement
+# the methods +.OPTIONS+, +#metadata+, and +#save+.
+module MiGA::Common::WithOption
+  def option(key)
+    assert_has_option(key)
+    opt = option_by_metadata(key)
+    value = opt.nil? ? option_by_default(key) : opt
+    value = value[self] if value.is_a?(Proc)
+    value
+  end
+
+  def set_option(key, value, from_string = false)
+    metadata[key] = assert_valid_option_value(key, value, from_string)
+    save
+    option(key)
+  end
+
+  def all_options
+    Hash[self.class.OPTIONS.each_key.map { |key| [key, option(key)] }]
+  end
+
+  def option?(key)
+    !self.class.OPTIONS[key.to_sym].nil?
+  end
+
+  def option_by_metadata(key)
+    metadata[key]
+  end
+
+  def option_by_default(key)
+    self.class.OPTIONS[key.to_sym][:default]
+  end
+
+  def assert_has_option(key)
+    opt = self.class.OPTIONS[key.to_sym]
+    raise "Unrecognized option: #{key}" if opt.nil?
+    opt
+  end
+
+  def assert_valid_option_value(key, value, from_string = false)
+    opt = assert_has_option(key)
+    value = option_from_string(key, value) if from_string
+
+    # nil is always valid, and so are supported tokens
+    return value if value.nil? || opt[:tokens]&.include?(value)
+
+    if opt[:type] && !value.is_a?(opt[:type])
+      raise "Invalid value type for #{key}: #{value.class}, not #{opt[:type]}"
+    end
+
+    if opt[:in] && !opt[:in].include?(value)
+      raise "Value out of range for #{key}: #{value}, not #{opt[:in]}"
+    end
+
+    value
+  end
+
+  def option_from_string(key, value)
+    opt = assert_has_option(key)
+
+    if ['', 'nil'].include?(value)
+      nil
+    elsif opt[:tokens]&.include?(value)
+      value
+    elsif opt[:type]&.equal?(Float)
+      raise "Not a float: #{value}" unless value =~ /^-?\.?\d/
+      value.to_f
+    elsif opt[:type]&.equal?(Integer)
+      raise "Not an integer: #{value}" unless value =~ /^-?\d/
+      value.to_i
+    elsif opt[:in]&.include?(true) && value == 'true'
+      true
+    elsif opt[:in]&.include?(false) && value == 'false'
+      false
+    else
+      value
+    end
+  end
+end

data/lib/miga/common/with_result.rb

@@ -86,7 +86,8 @@ module MiGA::Common::WithResult
         if res.nil?
           # Run if the step has not been calculated,
           # unless too many attempts were already made
-          if (metadata["_try_#{t}"] || 0) > (project.metadata[:max_try] || 10)
+          cur_try = metadata["_try_#{t}"] || 0
+          if cur_try > project.option(:max_try)
             inactivate! "Too many errors in step #{t}"
             false
           else
@@ -103,7 +104,7 @@ module MiGA::Common::WithResult
   ##
   # Mark all results for recalculation
   def recalculate_tasks(reason = nil)
-    each_result { |res| res.recalculate!(reason).save }
+    each_result { |_k, res| res.recalculate!(reason).save }
   end
 
 end

data/lib/miga/dataset/base.rb

@@ -1,7 +1,10 @@
-# @package MiGA
-# @license Artistic-2.0
+# frozen_string_literal: true
+
+require 'miga/common/with_option'
 
 class MiGA::Dataset < MiGA::MiGA
+  include MiGA::Common::WithOption
+
   # Class-level
   class << self
     def RESULT_DIRS
@@ -15,6 +18,10 @@ class MiGA::Dataset < MiGA::MiGA
     def PREPROCESSING_TASKS
       @@PREPROCESSING_TASKS
     end
+
+    def OPTIONS
+      @@OPTIONS
+    end
   end
 end
 
@@ -85,4 +92,15 @@ module MiGA::Dataset::Base
   # tasks are ignored for single-organism datasets or for unknwon types.
   @@ONLY_MULTI_TASKS = [:mytaxa]
   @@_ONLY_MULTI_TASKS_H = Hash[@@ONLY_MULTI_TASKS.map { |i| [i, true] }]
+
+  ##
+  # Options supported by datasets
+  @@OPTIONS = {
+    db_project: {
+      desc: 'Project to use as database', type: String
+    },
+    dist_req: {
+      desc: 'Run distances against these datasets', type: Array, default: []
+    }
+  }
 end

data/lib/miga/dataset/result.rb

@@ -50,7 +50,7 @@ module MiGA::Dataset::Result
       :upstream
     elsif !metadata["run_#{task}"].nil?
       metadata["run_#{task}"] ? :execute : :force
-    elsif task == :taxonomy && project.metadata[:ref_project].nil?
+    elsif task == :taxonomy && project.option(:ref_project).nil?
       :project
     elsif @@_EXCLUDE_NOREF_TASKS_H[task] && !ref?
       :noref
@@ -290,7 +290,8 @@ module MiGA::Dataset::Result
       MiGA::Result.new("#{base}.json"), name,
       longest_ssu_gene: '.ssu.fa',
       gff: '.ssu.gff',
-      all_ssu_genes: '.ssu.all.fa'
+      all_ssu_genes: '.ssu.all.fa',
+      classification: '.rdp.tsv'
     )
     opts[:is_clean] ||= false
     r.clean! if opts[:is_clean]

data/lib/miga/metadata.rb

@@ -56,24 +56,20 @@ class MiGA::Metadata < MiGA::MiGA
   ##
   # Save the metadata into #path
   def save
-    MiGA.DEBUG "Metadata.save #{path}"
+    return if self[:never_save]
+
+    MiGA::MiGA.DEBUG "Metadata.save #{path}"
     self[:updated] = Time.now.to_s
     json = to_json
-    sleeper = 0.0
-    slept = 0
-    while File.exist?(lock_file)
-      MiGA::MiGA.DEBUG "Waiting for lock: #{lock_file}"
-      sleeper += 0.1 if sleeper <= 10.0
-      sleep(sleeper.to_i)
-      slept += sleeper.to_i
-      raise "Lock detected for over 10 minutes: #{lock_file}" if slept > 600
-    end
-    FileUtils.touch lock_file
+    wait_for_lock
+    FileUtils.touch(lock_file)
     ofh = File.open("#{path}.tmp", 'w')
     ofh.puts json
     ofh.close
-    raise "Lock-racing detected for #{path}" unless
-      File.exist?("#{path}.tmp") and File.exist?(lock_file)
+
+    unless File.exist?("#{path}.tmp") && File.exist?(lock_file)
+      raise "Lock-racing detected for #{path}"
+    end
 
     File.rename("#{path}.tmp", path)
     File.unlink(lock_file)
@@ -154,4 +150,20 @@ class MiGA::Metadata < MiGA::MiGA
   def to_json
     MiGA::Json.generate(data)
   end
+
+  private
+
+  ##
+  # Wait for the lock to go away
+  def wait_for_lock
+    sleeper = 0.0
+    slept = 0.0
+    while File.exist?(lock_file)
+      MiGA::MiGA.DEBUG "Waiting for lock: #{lock_file}"
+      sleeper += 0.1 if sleeper <= 10.0
+      sleep(sleeper)
+      slept += sleeper
+      raise "Lock detected for over 10 minutes: #{lock_file}" if slept > 600
+    end
+  end
 end

data/lib/miga/project/base.rb

@@ -1,7 +1,10 @@
-# @package MiGA
-# @license Artistic-2.0
+# frozen_string_literal: true
+
+require 'miga/common/with_option'
 
 class MiGA::Project < MiGA::MiGA
+  include MiGA::Common::WithOption
+
   class << self
     ##
     # Does the project at +path+ exist?
@@ -33,6 +36,10 @@ class MiGA::Project < MiGA::MiGA
     def RESULT_DIRS
       @@RESULT_DIRS
     end
+
+    def OPTIONS
+      @@OPTIONS
+    end
   end
 end
 
@@ -108,4 +115,77 @@ module MiGA::Project::Base
   ##
   # Project-wide tasks for :clade projects
   @@INCLADE_TASKS = [:subclades, :ogs]
+
+  ##
+  # Options supported by projects
+  @@OPTIONS = {
+    ref_project: {
+      desc: 'Project with reference taxonomy', type: String
+    },
+    db_proj_dir: {
+      desc: 'Directory containing database projects', type: String
+    },
+    tax_pvalue: {
+      desc: 'Maximum p-value to transfer taxonomy', default: 0.05, type: Float,
+      in: 0.0..1.0
+    },
+    haai_p: {
+      desc: 'Value of aai.rb -p on hAAI', type: String,
+      default: proc { |project| project.clade? ? 'no' : 'blast+' },
+      in: %w[blast+ blast blat diamond no]
+    },
+    aai_p: {
+      desc: 'Value of aai.rb -p on AAI', default: 'blast+', type: String,
+      in: %w[blast+ blast blat diamond]
+    },
+    ani_p: {
+      desc: 'Value of ani.rb -p on ANI', default: 'blast+', type: String,
+      in: %w[blast+ blast blat fastani]
+    },
+    max_try: {
+      desc: 'Maximum number of task attempts', default: 10, type: Integer,
+      in: (0..1000)
+    },
+    aai_save_rbm: {
+      desc: 'Should RBMs be saved for OGS analysis?',
+      default: proc { |project| project.clade? },
+      in: [true, false]
+    },
+    ogs_identity: {
+      desc: 'Min RBM identity for OGS', default: 80.0, type: Float,
+      in: (0.0..100.0)
+    },
+    clean_ogs: {
+      desc: 'If false, keeps ABC files (clades only)', default: true,
+      in: [true, false]
+    },
+    run_clades: {
+      desc: 'Should clades be estimated from distances?', default: true,
+      in: [true, false]
+    },
+    gsp_ani: {
+      desc: 'ANI limit to propose gsp clades', default: 95.0, type: Float,
+      in: (0.0..100.0)
+    },
+    gsp_aai: {
+      desc: 'AAI limit to propose gsp clades', default: 90.0, type: Float,
+      in: (0.0..100.0)
+    },
+    gsp_metric: {
+      desc: 'Metric to propose clades', default: 'ani', type: String,
+      in: %w[ani aai]
+    },
+    ess_coll: {
+      desc: 'Collection of essential genes to use', default: 'dupont_2012',
+      type: String, in: %w[dupont_2012 lee_2019]
+    },
+    min_qual: {
+      desc: 'Minimum genome quality', default: 25.0, type: Float,
+      in: -Float::INFINITY..100.0, tokens: %w[no]
+    },
+    distances_checkpoint: {
+      desc: 'Number of comparisons before storing data', default: 10,
+      type: Integer, in: 1...Float::INFINITY
+    }
+  }
 end

data/lib/miga/project/result.rb

@@ -31,9 +31,9 @@ module MiGA::Project::Result
   ##
   # Is this +task+ to be bypassed?
   def ignore_task?(task)
-    metadata["run_#{task}"] == false ||
-      (!is_clade? && @@INCLADE_TASKS.include?(task) &&
-        metadata["run_#{task}"] != true)
+    return true if metadata["run_#{task}"] == false
+
+    !clade? && @@INCLADE_TASKS.include?(task) && metadata["run_#{task}"] != true
   end
 
   ##
@@ -74,7 +74,7 @@ module MiGA::Project::Result
       return r
     end
     return nil unless result_files_exist?(base, %w[.proposed-clades])
-    unless is_clade? ||
+    unless clade? ||
            result_files_exist?(
              base, %w[.pdf .classif .medoids .class.tsv .class.nwk]
            )

data/lib/miga/remote_dataset.rb

@@ -162,6 +162,8 @@ class MiGA::RemoteDataset < MiGA::MiGA
       txt.empty? ? sleep(1) : break
     end
     doc = MiGA::Json.parse(txt, symbolize: false, contents: true)
+    return if doc.nil? || doc['result'].nil? || doc['result'].empty?
+
     @_ncbi_asm_json_doc = doc['result'][ doc['result']['uids'].first ]
   end
 

data/lib/miga/result/stats.rb

@@ -118,7 +118,7 @@ module MiGA::Result::Stats
 
   def compute_stats_essential_genes
     stats = {}
-    if source.is_multi?
+    if source.multi?
       stats = { median_copies: 0, mean_copies: 0 }
       File.open(file_path(:report), 'r') do |fh|
         fh.each_line do |ln|
@@ -151,7 +151,7 @@ module MiGA::Result::Stats
       source.save
 
       # Inactivate low-quality datasets
-      min_qual = (project.metadata[:min_qual] || 25)
+      min_qual = project.option(:min_qual)
       if min_qual != 'no' && stats[:quality] < min_qual
         source.inactivate! 'Low quality genome'
       end

data/lib/miga/version.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'date'
 
 ##
@@ -8,7 +10,7 @@ module MiGA
   # - Float representing the major.minor version.
   # - Integer representing gem releases of the current version.
   # - Integer representing minor changes that require new version number.
-  VERSION = [0.7, 21, 0]
+  VERSION = [0.7, 24, 0].freeze
 
   ##
   # Nickname for the current major.minor version.
@@ -16,7 +18,7 @@ module MiGA
 
   ##
   # Date of the current gem release.
-  VERSION_DATE = Date.new(2021, 1, 13)
+  VERSION_DATE = Date.new(2021, 2, 16)
 
   ##
   # Reference of MiGA.