miga-base 0.7.23.0 → 0.7.25.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 6ed8981c0dfc266aea53be66e06b77eaf277444a094ef460021c2713aef3bf73
-  data.tar.gz: d9f0bd3934dd5e3ba27cb58adfd1236a74ba6b003a06cd74b9da96e594a27e99
+  metadata.gz: 47313defda91b1a031e7700e193fadf688a0115deedf77e40013c16e79bb8b49
+  data.tar.gz: ab6ea787eee6aed1c28ca09a97c8d580b23985fb3961dbf3fa7a33fb2a14f4e5
 SHA512:
-  metadata.gz: f40242b96f36ce5528e2206aa4f366d047b5969a45d69cff5ac825297563f9adddaee957e562eddae34c96b8a0651fd2fec39a42d2cd278d66ea03ac35afabfc
-  data.tar.gz: ab01b09ab020c1bfcf5dd5be7a01c971e83e02066ad35f4b6a9f0c525351c81b5cdb0e80910cf4237b9e59a33c518995a073c1f625f049b6b722802541d11c1f
+  metadata.gz: b152125992f0ece34a0a1d59ccd3d69ed5c931c5fada4a167a12bf3e53054505bde9974fc7e43d32c1164a8d9f89b69135638d71c572b015b04a311bb38540a4
+  data.tar.gz: 9a5493e2d025a4c245ed9b33a2059f529b6ff265274685d602869180538895807034cb24d1628a2f7ed2a9e9124f9b6097d49ad845f736c7c05355cc4bcf7add

data/Gemfile

@@ -1,6 +1,9 @@
 source "https://rubygems.org"
 gemspec name: "miga-base"
+
 group :test do
   gem "simplecov"
   gem "codeclimate-test-reporter", "~> 1.0.0"
 end
+
+gem "rake", "~> 12.0"

data/Rakefile

@@ -1,3 +1,4 @@
+require 'bundler/gem_tasks'
 require 'rake/testtask'
 
 SOURCES = FileList['lib/**/*.rb']

data/lib/miga/cli/action/add.rb

@@ -6,10 +6,7 @@ require 'miga/cli/action'
 class MiGA::Cli::Action::Add < MiGA::Cli::Action
   def parse_cli
     cli.expect_files = true
-    cli.defaults = {
-      ref: true, ignore_dups: false,
-      regexp: MiGA::Cli.FILE_REGEXP
-    }
+    cli.defaults = { ref: true, ignore_dups: false }
     cli.parse do |opt|
       opt.separator 'You can create multiple datasets with a single command; ' \
         'simply pass all the files at the end: {FILES...}'
@@ -37,7 +34,10 @@ class MiGA::Cli::Action::Add < MiGA::Cli::Action
       opt.on(
         '-R', '--name-regexp REGEXP', Regexp,
         'Regular expression indicating how to extract the name from the path',
-        "By default: '#{cli[:regexp]}'"
+        'By default for paired files:',
+        "'#{MiGA::Cli.FILE_REGEXP(true)}'",
+        'By default for other files:',
+        "'#{MiGA::Cli.FILE_REGEXP}'"
       ) { |v| cli[:regexp] = v }
       opt.on(
         '--prefix STRING',
@@ -59,14 +59,16 @@ class MiGA::Cli::Action::Add < MiGA::Cli::Action
     p = cli.load_project
     files, file_type = get_files_and_type
 
+    paired = cli[:input_type].to_s.include?('_paired')
+    cli[:regexp] ||= MiGA::Cli.FILE_REGEXP(paired)
+
     cli.say 'Creating datasets:'
     files.each do |file|
       d = create_dataset(file, p)
       next if d.nil?
 
       copy_file_to_project(file, file_type, d, p)
-      d = cli.add_metadata(d)
-      d.save
+      cli.add_metadata(d)
       p.add_dataset(d.name)
       res = d.first_preprocessing(true)
       cli.say "  result: #{res}"
@@ -167,7 +169,7 @@ class MiGA::Cli::Action::Add < MiGA::Cli::Action
     file_type[2].each_with_index do |ext, i|
       gz = file[i] =~ /\.gz/ ? '.gz' : ''
       FileUtils.cp(file[i], "#{r_path}#{ext}#{gz}")
-      cli.say "  file: #{file[i]}"
+      cli.say "  file: #{File.basename(file[i])}"
     end
     File.open("#{r_path}.done", 'w') { |f| f.print Time.now.to_s }
   end

data/lib/miga/cli/action/classify_wf.rb

@@ -49,23 +49,24 @@ class MiGA::Cli::Action::ClassifyWf < MiGA::Cli::Action
   def perform
     # Input data
     ref_db = reference_db
-    p_metadata = Hash[
-      %w[project_stats haai_distances aai_distances ani_distances clade_finding]
-                 .map { |i| ["run_#{i}", false] }
+    norun = %w[
+      project_stats haai_distances aai_distances ani_distances clade_finding
     ]
-    p_metadata[:ref_project] = ref_db.path
-    p_metadata[:tax_pvalue] = cli[:pvalue]
-    p = create_project(:assembly, p_metadata,
-                       run_ssu: false, run_mytaxa_scan: false, run_distances: false)
+    p_metadata = Hash[norun.map { |i| ["run_#{i}", false] }]
+    p = create_project(
+      :assembly,
+      p_metadata,
+      run_ssu: false, run_mytaxa_scan: false, run_distances: false
+    )
+    p.set_option(:ref_project, ref_db.path)
+    p.set_option(:tax_pvalue, cli[:pvalue], true)
     # Run
     run_daemon
     summarize(%w[cds assembly essential_genes]) if cli[:summaries]
     summarize(['taxonomy'])
     cli.say "Summary: classification"
-    call_cli([
-               'ls', '-P', cli[:outdir], '-m', 'tax', '--tab',
-               '-o', File.expand_path('classification.tsv', cli[:outdir])
-             ])
+    ofile = File.expand_path('classification.tsv', cli[:outdir])
+    call_cli(['ls', '-P', cli[:outdir], '-m', 'tax', '--tab', '-o', ofile])
     cleanup
   end
 

data/lib/miga/cli/action/derep_wf.rb

@@ -52,17 +52,11 @@ class MiGA::Cli::Action::DerepWf < MiGA::Cli::Action
     # Input data
     p = create_project(
       :assembly,
-      {
-        run_project_stats: false,
-        run_clades: false,
-        gsp_metric: cli[:metric],
-        :"gsp_#{cli[:metric]}" => cli[:threshold]
-      },
+      { run_project_stats: false, run_clades: false },
       { run_mytaxa_scan: false, run_ssu: false }
     )
-    unless cli[:threshold] >= 0.0 && cli[:threshold] <= 100.0
-      raise 'The threshold of identity must be in the range [0,100]'
-    end
+    p.set_option(:gsp_metric, cli[:metric].to_s)
+    p.set_option(:"gsp_#{cli[:metric]}", cli[:threshold])
 
     # Run
     run_daemon

data/lib/miga/cli/action/edit.rb

@@ -34,6 +34,5 @@ class MiGA::Cli::Action::Edit < MiGA::Cli::Action
       cli[:activate] ? obj.activate! : obj.inactivate!(cli[:reason])
     end
     cli.add_metadata(obj)
-    obj.save
   end
 end

data/lib/miga/cli/action/find.rb

@@ -37,7 +37,7 @@ class MiGA::Cli::Action::Find < MiGA::Cli::Action
       if cli[:add]
         cli.say "Registering: #{dn}"
         d = Dataset.new(p, dn, cli[:ref])
-        d = add_metadata(d)
+        add_metadata(d)
         p.add_dataset(dn)
         res = d.first_preprocessing(true)
         cli.say "- #{res}"

data/lib/miga/cli/action/generic.rb

@@ -20,7 +20,7 @@ class MiGA::Cli::Action::Generic < MiGA::Cli::Action
       opt.on(
         '-v', '--version',
         'Show MiGA version'
-      ) { puts MiGA::MiGA.VERSION; exit }
+      ) { puts MiGA::MiGA.FULL_VERSION; exit }
       opt.on(
         '-V', '--long-version',
         'Show complete MiGA version'

data/lib/miga/cli/action/get.rb

@@ -6,8 +6,9 @@ require 'miga/remote_dataset'
 
 class MiGA::Cli::Action::Get < MiGA::Cli::Action
   def parse_cli
-    cli.defaults = { query: false, universe: :ncbi, db: :nuccore,
-                     get_md: false, only_md: false }
+    cli.defaults = {
+      query: false, universe: :ncbi, db: :nuccore, get_md: false, only_md: false
+    }
     cli.parse do |opt|
       cli.opt_object(opt, [:project, :dataset, :dataset_type])
       opt.on(
@@ -141,6 +142,10 @@ class MiGA::Cli::Action::Get < MiGA::Cli::Action
 
   def create_dataset(sub_cli, p, rd)
     sub_cli.say 'Creating dataset'
+    if Dataset.exist?(p, sub_cli[:dataset])
+      raise "Dataset already exists: #{sub_cli[:dataset]}"
+    end
+
     dummy_d = Dataset.new(p, sub_cli[:dataset])
     md = sub_cli.add_metadata(dummy_d).metadata.data
     md[:metadata_only] = true if cli[:only_md]

data/lib/miga/cli/action/index_wf.rb

@@ -16,8 +16,10 @@ class MiGA::Cli::Action::IndexWf < MiGA::Cli::Action
         'Perform MyTaxa scan analysis'
       ) { |v| cli[:mytaxa] = v }
       opts_for_wf_distances(opt)
-      opts_for_wf(opt, 'Input genome assemblies (nucleotides, FastA)',
-                  cleanup: false, project_type: true)
+      opts_for_wf(
+        opt, 'Input genome assemblies (nucleotides, FastA)',
+        cleanup: false, project_type: true
+      )
     end
   end
 

data/lib/miga/cli/action/init.rb

@@ -61,8 +61,7 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
     check_configuration_script(rc_fh)
     paths = check_software_requirements(rc_fh)
     check_additional_files(paths)
-    check_r_packages(paths)
-    check_ruby_gems(paths)
+    check_libraries(paths)
     configure_daemon
     close_rc_file(rc_fh)
     cli.puts 'Configuration complete. MiGA is ready to work!'
@@ -83,34 +82,6 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
     )
   end
 
-  def test_r_package(cli, paths, pkg)
-    run_r_cmd(cli, paths, "library('#{pkg}')")
-    $?.success?
-  end
-
-  def install_r_package(cli, paths, pkg)
-    r_cmd = "install.packages('#{pkg}', repos='http://cran.rstudio.com/')"
-    run_r_cmd(cli, paths, r_cmd)
-  end
-
-  def test_ruby_gem(cli, paths, pkg)
-    run_cmd(
-      cli,
-      "#{paths['ruby'].shellescape} -r #{pkg.shellescape} -e '' 2>/dev/null"
-    )
-    $?.success?
-  end
-
-  def install_ruby_gem(cli, paths, pkg)
-    gem_cmd = "Gem::GemRunner.new.run %w(install --user #{pkg})"
-    run_cmd(
-      cli,
-      "#{paths['ruby'].shellescape} \
-        -r rubygems -r rubygems/gem_runner \
-        -e #{gem_cmd.shellescape} 2>&1"
-    )
-  end
-
   def list_requirements
     if cli.ask_user(
       'Would you like to see all the requirements before starting?',
@@ -189,40 +160,70 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
     path
   end
 
-  def check_r_packages(paths)
-    cli.puts 'Looking for R packages:'
-    %w(ape cluster vegan).each do |pkg|
-      cli.print "Testing #{pkg}... "
-      if test_r_package(cli, paths, pkg)
-        cli.puts 'yes'
-      else
-        cli.puts 'no, installing'
-        cli.print '' + install_r_package(cli, paths, pkg)
-        unless test_r_package(cli, paths, pkg)
-          raise "Unable to auto-install R package: #{pkg}"
-        end
+  def check_libraries(paths)
+    req_libraries = {
+      r: %w[ape cluster vegan],
+      ruby: %w[sqlite3 daemons json],
+      python: %w[numpy]
+    }
+
+    req_libraries.each do |language, libraries|
+      cli.puts "Looking for #{language.to_s.capitalize} libraries:"
+      libraries.each do |lib|
+        check_and_install_library(paths, language, lib)
       end
+      cli.puts ''
     end
-    cli.puts ''
   end
 
-  def check_ruby_gems(paths)
-    cli.puts 'Looking for Ruby gems:'
-    %w(sqlite3 daemons json).each do |pkg|
-      cli.print "Testing #{pkg}... "
-      if test_ruby_gem(cli, paths, pkg)
-        cli.puts 'yes'
-      else
-        cli.puts 'no, installing'
-        # This hackey mess is meant to ensure the test and installation are done
-        # on the configuration Ruby, not on the Ruby currently executing the
-        # init action
-        cli.print install_ruby_gem(cli, paths, pkg)
-        unless test_ruby_gem(cli, paths, pkg)
-          raise "Unable to auto-install Ruby gem: #{pkg}"
-        end
+  def check_and_install_library(paths, language, library)
+    cli.print "Testing #{library}... "
+    if test_library(cli, paths, language, library)
+      cli.puts 'yes'
+    else
+      cli.puts 'no, installing'
+      cli.print '' + install_library(cli, paths, language, library)
+      unless test_library(cli, paths, language, library)
+        raise "Cannot install #{language.to_s.capitalize} library: #{library}"
       end
     end
-    cli.puts ''
+  end
+
+  def test_library(cli, paths, language, pkg)
+    case language
+    when :r
+      run_r_cmd(cli, paths, "library('#{pkg}')")
+    when :ruby
+      x = "#{paths['ruby'].shellescape} -r #{pkg.shellescape} -e '' 2>/dev/null"
+      run_cmd(cli, x)
+    when :python
+      x = "#{paths['python3'].shellescape} -c 'import #{pkg}' 2>/dev/null"
+      run_cmd(cli, x)
+    else
+      raise "Unrecognized language: #{language}"
+    end
+    $?.success?
+  end
+
+  def install_library(cli, paths, language, pkg)
+    case language
+    when :r
+      r_cmd = "install.packages('#{pkg}', repos='http://cran.rstudio.com/')"
+      run_r_cmd(cli, paths, r_cmd)
+    when :ruby
+      # This hackey mess is meant to ensure the test and installation are done
+      # on the configuration Ruby, not on the Ruby currently executing the
+      # init action
+      gem_cmd = "Gem::GemRunner.new.run %w(install --user #{pkg})"
+      x = "#{paths['ruby'].shellescape} -r rubygems -r rubygems/gem_runner \
+            -e #{gem_cmd.shellescape} 2>&1"
+      run_cmd(cli, x)
+    when :python
+      x = "#{paths['python3'].shellescape} \
+            -m pip install --user #{pkg.shellescape} 2>&1"
+      run_cmd(cli, x)
+    else
+      raise "Unrecognized language: #{language}"
+    end
   end
 end

data/lib/miga/cli/action/init/files_helper.rb

@@ -26,7 +26,7 @@ module MiGA::Cli::Action::Init::FilesHelper
   def close_rc_file(rc_fh)
     rc_fh.puts <<~FOOT
 
-      MIGA_CONFIG_VERSION='#{MiGA::MiGA.VERSION}'
+      MIGA_CONFIG_VERSION='#{MiGA::MiGA.FULL_VERSION}'
       MIGA_CONFIG_LONGVERSION='#{MiGA::MiGA.LONG_VERSION}'
       MIGA_CONFIG_DATE='#{Time.now}'
 
@@ -59,6 +59,7 @@ module MiGA::Cli::Action::Init::FilesHelper
     end
     check_rdp_classifier if cli[:rdp]
     check_phyla_lite
+    cli.puts ''
   end
 
   def check_mytaxa_scores(paths)

data/lib/miga/cli/action/ncbi_get.rb

@@ -269,7 +269,7 @@ class MiGA::Cli::Action::NcbiGet < MiGA::Cli::Action
     else
       cli.say '  Creating dataset'
       rd.save_to(p, name, !cli[:query], body[:md])
-      cli.add_metadata(p.add_dataset(name)).save
+      cli.add_metadata(p.add_dataset(name))
     end
   end
 end

data/lib/miga/cli/action/new.rb

@@ -24,6 +24,11 @@ class MiGA::Cli::Action::New < MiGA::Cli::Action
         'Use faster identity engines (Diamond-AAI and FastANI)',
         'Equivalent to: -m aai_p=diamond,ani_p=fastani'
       ) { |v| cli[:fast] = v }
+      opt.on(
+        '--sensitive',
+        'Use more sensitive identity engines (BLAST+)',
+        'Equivalent to: -m aai_p=blast+,ani_p=blast+'
+      ) { |v| cli[:sensitive] = v }
       opt.on(
         '-m', '--metadata STRING',
         'Metadata as key-value pairs separated by = and delimited by comma',
@@ -35,20 +40,21 @@ class MiGA::Cli::Action::New < MiGA::Cli::Action
   def perform
     cli.ensure_type(MiGA::Project)
     cli.ensure_par(project: '-P')
-    unless File.exist?(File.join(ENV['HOME'], '.miga_rc')) &&
-           File.exist?(File.join(ENV['HOME'], '.miga_daemon.json'))
-      raise "You must initialize MiGA before creating the first project.\n" +
-            'Please use "miga init".'
+    unless MiGA::MiGA.initialized?
+      raise 'MiGA has not been initialized, please use "miga init" first'
     end
     cli.say "Creating project: #{cli[:project]}"
-    raise 'Project already exists, aborting.' if Project.exist? cli[:project]
+    raise 'Project already exists, aborting' if Project.exist?(cli[:project])
 
     p = Project.new(cli[:project], false)
     p = cli.add_metadata(p)
-    if cli[:fast]
-      p.metadata[:aai_p] = 'diamond'
-      p.metadata[:ani_p] = 'fastani'
+
+    if cli[:sensitive]
+      p.set_option(:aai_p, 'blast+')
+      p.set_option(:ani_p, 'blast+')
+    elsif cli[:fast]
+      p.set_option(:aai_p, 'diamond')
+      p.set_option(:ani_p, 'fastani')
     end
-    p.save
   end
 end

data/lib/miga/cli/action/option.rb

@@ -0,0 +1,63 @@
+# frozen_string_literal: true
+
+require 'miga/cli/action'
+
+##
+# CLI: `miga option`
+class MiGA::Cli::Action::Option < MiGA::Cli::Action
+  def parse_cli
+    cli.parse do |opt|
+      cli.opt_object(opt, %i[project dataset_opt])
+      opt.on(
+        '-k', '--key STRING',
+        'Option name to get or set (by default, all options are printed)'
+      ) { |v| cli[:key] = v }
+      opt.on(
+        '--value STRING',
+        'Value of the option to set (by default, option value is not changed)',
+        'Recognized tokens: nil, true, false'
+      ) { |v| cli[:value] = v }
+      opt.on(
+        '--about',
+        'Print additional information about the values supported by this option'
+      ) { |v| cli[:about] = v }
+      opt.on(
+        '--tab',
+        'Return a tab-delimited table'
+      ) { |v| cli[:tabular] = v }
+      opt.on(
+        '-o', '--output PATH',
+        'Create output file instead of returning to STDOUT'
+      ) { |v| cli[:output] = v }
+    end
+  end
+
+  def perform
+    unless cli[:value].nil?
+      cli.ensure_par(
+        { key: '-k' },
+        '%<name>s is mandatory when --value is set: provide %<flag>s'
+      )
+    end
+    obj = cli.load_project_or_dataset
+    io = cli[:output].nil? ? $stdout : File.open(cli[:output], 'w')
+    if cli[:key].nil?
+      opts = obj.all_options
+                .map { |k, v| [k, v, obj.assert_has_option(k)[:desc]] }
+      cli.table(%w[Key Value Definition], opts, io)
+    elsif cli[:about]
+      opt = obj.assert_has_option(cli[:key])
+      title = "#{cli[:key]}: #{opt[:desc]}"
+      io.puts title
+      io.puts '-' * title.length
+      opt.each do |k, v|
+        v = v[obj] if v.is_a? Proc
+        io.puts "#{k.to_s.capitalize}: #{v}" unless k == :desc
+      end
+    else
+      obj.set_option(cli[:key], cli[:value], true) unless cli[:value].nil?
+      io.puts obj.option(cli[:key])
+    end
+    io.close unless cli[:output].nil?
+  end
+end