miga-base 0.7.21.0 → 0.7.24.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: ed3a4b61cfc055c8f305438f304257116b7ec3fe92e10af4d00150d6835cf079
-  data.tar.gz: 9cad53f9a308fbb96d0264932348543339a3d03e096695217b2af85f5b2ac89f
+  metadata.gz: 82ca84b468df712ab38eba74071c26e8dcb1d87335764dde2ba86075abc8c8d5
+  data.tar.gz: 36583e298fc020faa5cc16af2f6a3fcc88c30619a4f529fd6f03772f6c437bb9
 SHA512:
-  metadata.gz: 1507b05bdc37fe6b8f7dc1b7531042d225dc2e79112b3fb9dcb1e4a558e38321a1a8a1203024cb6f956623be928c676fa3c9c1fb971190e2ef6e26328a97d5e5
-  data.tar.gz: 570bd7b58bc0a6262ef77b1be1a797783b423d76f474fa4840eb146fdbc5fefc12dc5d8abd34f72c4dc7a7cc8ad000fe80005c33af8c9dbcb4232a757f01e07a
+  metadata.gz: d78a3709fecba4594d8dc9a2351f217b3605022237e424eb1a194ccf62ddf2059eb1df274f68a15cda74936dfece3f2123485483773172cc62cfba9ff454c5d2
+  data.tar.gz: fbf11b04c062701b204f4ba93425dea6f8ff1492fe48f7182472e65b16fccbde98e86d396a79cf51c6e410a246c3b9e9509043eaf952b320191ec086e2efa573

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/README.md

@@ -1,6 +1,6 @@
 [![Code Climate](https://codeclimate.com/github/bio-miga/miga/badges/gpa.svg)](https://codeclimate.com/github/bio-miga/miga)
 [![Test Coverage](https://codeclimate.com/github/bio-miga/miga/badges/coverage.svg)](https://codeclimate.com/github/bio-miga/miga/coverage)
-[![Build Status](https://travis-ci.org/bio-miga/miga.svg?branch=master)](https://travis-ci.org/bio-miga/miga)
+[![Build Status](https://github.com/bio-miga/miga/workflows/build/badge.svg)](https://github.com/bio-miga/miga/actions?query=workflow:build)
 [![Gem Version](https://badge.fury.io/rb/miga-base.svg)](https://badge.fury.io/rb/miga-base)
 [![Inch docs](http://inch-ci.org/github/bio-miga/miga.svg)](http://inch-ci.org/github/bio-miga/miga)
 

data/Rakefile

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

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

@@ -65,8 +65,7 @@ class MiGA::Cli::Action::Add < MiGA::Cli::Action
       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}"

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])
+    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/get_db.rb

@@ -2,9 +2,7 @@
 # @license Artistic-2.0
 
 require 'miga/cli/action'
-require 'net/ftp'
 require 'digest/md5'
-require 'open-uri'
 
 class MiGA::Cli::Action::GetDb < MiGA::Cli::Action
   def parse_cli
@@ -12,7 +10,7 @@ class MiGA::Cli::Action::GetDb < MiGA::Cli::Action
       database: :recommended,
       version: :latest,
       local: File.expand_path('.miga_db', ENV['MIGA_HOME']),
-      host: 'ftp://microbial-genomes.org/db',
+      host: MiGA::MiGA.known_hosts(:miga_db),
       pb: true,
       overwrite: true
     }
@@ -50,6 +48,14 @@ class MiGA::Cli::Action::GetDb < MiGA::Cli::Action
   end
 
   def perform
+    # Quick check when the database is not an alias
+    dir = File.join(cli[:local], cli[:database].to_s)
+    if !cli[:overwrite] && Dir.exist?(dir)
+      cli.puts "Database exists: #{dir}"
+      return
+    end
+
+    # Remote manifest
     @ftp = remote_connection
     manif = remote_manifest(@ftp)
     cli.puts "# Host: #{manif[:host]}"
@@ -59,6 +65,8 @@ class MiGA::Cli::Action::GetDb < MiGA::Cli::Action
     list_versions(db) and return
     ver = version_requested(db)
     check_target and return
+
+    # Download and expand
     file = download_file(@ftp, ver[:path])
     check_digest(ver, file)
     unarchive(file)
@@ -78,27 +86,14 @@ class MiGA::Cli::Action::GetDb < MiGA::Cli::Action
 
   def remote_connection
     cli.say "Connecting to '#{cli[:host]}'"
-    uri = URI.parse(cli[:host])
-    raise 'Only FTP hosts are supported' unless uri.scheme == 'ftp'
-
-    ftp = Net::FTP.new(uri.host)
-    ftp.passive = true
-    ftp.login
-    ftp.chdir(uri.path)
-    ftp
+    MiGA::MiGA.remote_connection(cli[:host])
   end
 
   def download_file(ftp, path)
     cli.say "Downloading '#{path}'"
-    Dir.mkdir(cli[:local]) unless Dir.exist? cli[:local]
     file = File.expand_path(path, cli[:local])
-    filesize = ftp.size(path)
-    transferred = 0
-    ftp.getbinaryfile(path, file, 1024) do |data|
-      if cli[:pb]
-        transferred += data.size
-        cli.advance("#{path}:", transferred, filesize)
-      end
+    MiGA::MiGA.download_file_ftp(ftp, path, file) do |n, size|
+      cli.advance("#{path}:", n, size) if cli[:pb]
     end
     cli.print "\n" if cli[:pb]
     file
@@ -165,7 +160,7 @@ class MiGA::Cli::Action::GetDb < MiGA::Cli::Action
   def check_target
     return false if cli[:overwrite]
 
-    file = File.expand_path(cli[:database].to_s, cli[:local])
+    file = File.join(cli[:local], cli[:database].to_s)
     if Dir.exist? file
       warn "The target directory already exists: #{file}"
       true
@@ -195,7 +190,7 @@ class MiGA::Cli::Action::GetDb < MiGA::Cli::Action
 
   def unarchive(file)
     cli.say "Unarchiving #{file}"
-    `cd "#{cli[:local]}" && tar -zxf "#{file}"`
+    `cd "#{cli[:local]}" && tar -zxf "#{file}" && rm "#{file}"`
   end
 
   def register_database(manif, db, ver)

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

@@ -1,18 +1,24 @@
-# @package MiGA
-# @license Artistic-2.0
+# frozen_string_literal: true
 
 require 'miga/cli/action'
 require 'shellwords'
 
 class MiGA::Cli::Action::Init < MiGA::Cli::Action
   require 'miga/cli/action/init/daemon_helper'
+  require 'miga/cli/action/init/files_helper'
   include MiGA::Cli::Action::Init::DaemonHelper
+  include MiGA::Cli::Action::Init::FilesHelper
 
   def parse_cli
     cli.interactive = true
-    cli.defaults = { mytaxa: nil,
-                     config: File.expand_path('.miga_modules', ENV['HOME']),
-                     ask: false, auto: false, dtype: :bash }
+    cli.defaults = {
+      mytaxa: nil,
+      rdp: nil,
+      config: File.join(ENV['MIGA_HOME'], '.miga_modules'),
+      ask: false,
+      auto: false,
+      dtype: :bash
+    }
     cli.parse do |opt|
       opt.on(
         '-c', '--config PATH',
@@ -21,9 +27,14 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
       ) { |v| cli[:config] = v }
       opt.on(
         '--[no-]mytaxa',
-        'Should I try setting up MyTaxa its dependencies?',
+        'Should I try setting up MyTaxa and its dependencies?',
         'By default: interactive (true if --auto)'
       ) { |v| cli[:mytaxa] = v }
+      opt.on(
+        '--[no-]rdp',
+        'Should I try setting up the RDP classifier?',
+        'By default: interactive (true if --auto)'
+      ) { |v| cli[:rdp] = v }
       opt.on(
         '--daemon-type STRING',
         'Type of daemon launcher, one of: bash, ssh, qsub, msub, slurm',
@@ -47,13 +58,13 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
     BANNER
     list_requirements
     rc_fh = open_rc_file
-    check_configuration_script rc_fh
-    paths = check_software_requirements rc_fh
-    check_additional_files paths
-    check_r_packages paths
-    check_ruby_gems paths
+    check_configuration_script(rc_fh)
+    paths = check_software_requirements(rc_fh)
+    check_additional_files(paths)
+    check_r_packages(paths)
+    check_ruby_gems(paths)
     configure_daemon
-    close_rc_file rc_fh
+    close_rc_file(rc_fh)
     cli.puts 'Configuration complete. MiGA is ready to work!'
     cli.puts ''
   end
@@ -116,58 +127,18 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
 
   private
 
-  def open_rc_file
-    rc_path = File.expand_path('.miga_rc', ENV['HOME'])
-    if File.exist? rc_path
-      if cli.ask_user(
-        'I found a previous configuration. Do you want to continue?',
-        'yes', %w(yes no)
-      ) == 'no'
-        cli.puts 'OK, see you soon!'
-        exit(0)
-      end
-    end
-    rc_fh = File.open(rc_path, 'w')
-    rc_fh.puts <<~BASH
-      #!/bin/bash
-      # `miga init` made this on #{Time.now}
-
-    BASH
-    rc_fh
-  end
-
-  def check_configuration_script(rc_fh)
-    unless File.exist? cli[:config]
-      cli[:config] = cli.ask_user(
-        'Is there a script I need to load at startup?',
-        cli[:config]
-      )
-    end
-    if File.exist? cli[:config]
-      cli[:config] = File.expand_path(cli[:config])
-      cli.puts "Found bash configuration script: #{cli[:config]}"
-      rc_fh.puts "MIGA_STARTUP='#{cli[:config]}'"
-      rc_fh.puts '. "$MIGA_STARTUP"'
-    else
-      cli[:config] = '/dev/null'
-    end
-    cli.puts ''
-  end
-
   def check_software_requirements(rc_fh)
     cli.puts 'Looking for requirements:'
-    ask_for_mytaxa
-    rc_fh.puts 'export MIGA_MYTAXA="no"' unless cli[:mytaxa]
+    ask_for_optional(:mytaxa, 'MyTaxa')
+    rc_fh.puts "export MIGA_MYTAXA='#{cli[:mytaxa] ? 'yes' : 'no'}'"
+    ask_for_optional(:rdp, 'RDP classifier')
+    rc_fh.puts "export MIGA_RDP='#{cli[:rdp] ? 'yes' : 'no'}'"
     paths = {}
     rc_fh.puts 'MIGA_PATH=""'
     req_path = File.expand_path('utils/requirements.txt', MiGA.root_path)
     File.open(req_path, 'r') do |fh|
       fh.each_line do |ln|
-        next if $. < 3
-
-        r = ln.chomp.split(/\t+/)
-        next if r[0] =~ /\(opt\)$/ && !cli[:mytaxa]
-
+        r = define_software(ln) or next
         cli.print "Testing #{r[0]}#{" (#{r[3]})" if r[3]}... "
         path = find_software(r[1])
         paths[r[1]] = File.expand_path(r[1], path).shellescape
@@ -178,11 +149,20 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
     paths
   end
 
-  def ask_for_mytaxa
-    if cli[:mytaxa].nil?
-      cli[:mytaxa] =
+  def define_software(ln)
+    r = ln.chomp.split(/\t+/)
+    return if %w[Software --------].include?(r[0])
+    return if r[0] =~ /\(mytaxa\)$/ && !cli[:mytaxa]
+    return if r[0] =~ /\(rdp\)$/ && !cli[:rdp]
+
+    r
+  end
+
+  def ask_for_optional(symbol, name)
+    if cli[symbol].nil?
+      cli[symbol] =
         cli.ask_user(
-          'Should I include MyTaxa modules?',
+          "Should I include #{name} modules?",
           'yes', %w(yes no)
         ) == 'yes'
     end
@@ -209,27 +189,6 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
     path
   end
 
-  def check_additional_files(paths)
-    if cli[:mytaxa]
-      cli.puts 'Looking for MyTaxa databases:'
-      mt = File.dirname paths['MyTaxa']
-      cli.print 'Looking for scores... '
-      unless Dir.exist?(File.expand_path('db', mt))
-        cli.puts "no\nExecute 'python2 #{mt}/utils/download_db.py'"
-        exit(1)
-      end
-      cli.puts 'yes'
-      cli.print 'Looking for diamond db... '
-      unless File.exist?(File.expand_path('AllGenomes.faa.dmnd', mt))
-        cli.puts "no\nDownload " \
-          "'http://enve-omics.ce.gatech.edu/data/public_mytaxa/" \
-          "AllGenomes.faa.dmnd' into #{mt}"
-        exit(1)
-      end
-      cli.puts ''
-    end
-  end
-
   def check_r_packages(paths)
     cli.puts 'Looking for R packages:'
     %w(ape cluster vegan).each do |pkg|
@@ -266,15 +225,4 @@ class MiGA::Cli::Action::Init < MiGA::Cli::Action
     end
     cli.puts ''
   end
-
-  def close_rc_file(rc_fh)
-    rc_fh.puts <<~FOOT
-
-      MIGA_CONFIG_VERSION='#{MiGA::MiGA.VERSION}'
-      MIGA_CONFIG_LONGVERSION='#{MiGA::MiGA.LONG_VERSION}'
-      MIGA_CONFIG_DATE='#{Time.now}'
-
-    FOOT
-    rc_fh.close
-  end
 end

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

@@ -1,5 +1,4 @@
-# @package MiGA
-# @license Artistic-2.0
+# frozen_string_literal: true
 
 ##
 # Helper module with daemon configuration functions for MiGA::Cli::Action::Init