miga-base 0.7.9.0 → 0.7.10.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d699b0ec0f4b7097439d64083fec6f3187662f9bcfa5d9921aebb27cdfd1c263
-  data.tar.gz: 7ebffb85482969f3304e643a638fdc6f45439da8d2f11d9b8cffe27713c6e286
+  metadata.gz: 30d61eb5847a28c4d83a43e6e47ff0738bb819218e677b9aa43de158b441e0ae
+  data.tar.gz: 46338ae15353b71fb6e7eff390c8bb976d1c11fc296bb83f33aab8ffba1a3fa8
 SHA512:
-  metadata.gz: d6517ec344ff45a1ec79db56250ad0c15423b6eac0bf4cad07def7e17dde238c3c7064a70df9f50a2411ec3b2dae27756483ba92c192553cf8fdcc9104acf730
-  data.tar.gz: 8858ed1fed6861deff6137950c42c6f8a872c9406dc6ef3969676642f3c945de21395c4c5975e6885ede2059f9ea9c11620ec3c764774dc0f5deeb9ac058bb7f
+  metadata.gz: 488e7888039bb9e08e7c257fdd7cb0cf34340766f73b9b46c28b332072f5f207f5dcfb0df08c98b27f5640158d3a14ae6d0f5ab19d78f3e9e1ef44e381d34e74
+  data.tar.gz: 992cfc225eeb2f4b8017260e66c67bc9df39a79757ecdb65e01766079e103385259c1eeb447fdc5e6b769990f96256d384dc4adddf184693a3bc94b8323b9a5a

data/README.md

@@ -41,6 +41,7 @@ Developed and maintained by [Luis M. Rodriguez-R][lrr]. MiGA is the result of a
 collaboration between [Kostas Lab][kostas] at the Georgia Institute of
 Technology and [RDP][rdp] at Michigan State University.
 
+See also the [complete list of contributors](manual/part1/contributors.md).
 
 # License
 

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

@@ -42,7 +42,9 @@ class MiGA::Cli::Action::ClassifyWf < MiGA::Cli::Action
         '--no-summaries',
         'Do not generate intermediate step summaries'
       ) { |v| cli[:summaries] = v }
-      opts_for_wf(opt, 'Input genome assemblies (nucleotides, FastA)')
+      opts_for_wf(
+        opt, 'Input genome assemblies (nucleotides, FastA)', qual: false
+      )
     end
   end
 

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

@@ -17,18 +17,21 @@ class MiGA::Cli::Action::Edit < MiGA::Cli::Action
         'Activate dataset; requires -D'
       ) { |v| cli[:activate] = v }
       opt.on(
-        '--inactivate',
-        'Inactivate dataset; requires -D'
-      ) { |v| cli[:activate] = !v }
+        '--inactivate [reason]',
+        'Inactivate dataset; requires -D',
+        'The argument is optional: reason to inactivate dataset'
+      ) { |v| cli[:activate] = false ; cli[:reason] = v }
     end
   end
 
   def perform
     obj = cli.load_project_or_dataset
     unless cli[:activate].nil?
-      cli.ensure_par({ dataset: '-D' },
-                     '%<name>s is mandatory with --[in-]activate: please provide %<flag>s')
-      cli[:activate] ? obj.activate! : obj.inactivate!
+      cli.ensure_par(
+        { dataset: '-D' },
+        '%<name>s is mandatory with --[in-]activate: please provide %<flag>s'
+      )
+      cli[:activate] ? obj.activate! : obj.inactivate!(cli[:reason])
     end
     cli.add_metadata(obj)
     obj.save

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

@@ -15,7 +15,10 @@ class MiGA::Cli::Action::QualityWf < MiGA::Cli::Action
         '-m', '--mytaxa-scan',
         'Perform MyTaxa scan analysis'
       ) { |v| cli[:mytaxa] = v }
-      opts_for_wf(opt, 'Input genome assemblies (nucleotides, FastA)')
+      opts_for_wf(
+        opt, 'Input genome assemblies (nucleotides, FastA)',
+        qual: false
+      )
     end
   end
 

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

@@ -15,7 +15,7 @@ module MiGA::Cli::Action::Wf
 
   def opts_for_wf(opt, files_desc, params = {})
     {
-      multi: false, cleanup: true, project_type: false, ncbi: true
+      multi: false, cleanup: true, project_type: false, ncbi: true, qual: true
     }.each { |k, v| params[k] = v if params[k].nil? }
     opt.on(
       '-o', '--out_dir PATH',
@@ -40,6 +40,13 @@ module MiGA::Cli::Action::Wf
         'Only download complete genomes, not drafts'
       ) { |v| cli[:ncbi_draft] = v }
     end
+    if params[:qual]
+      opt.on(
+        '--min-qual FLOAT', Float,
+        'Minimum genome quality to include in analysis',
+        'By default: 50.0'
+      ) { |v| cli[:min_qual] = v }
+    end
     if params[:cleanup]
       opt.on(
         '-c', '--clean',
@@ -125,7 +132,7 @@ module MiGA::Cli::Action::Wf
              ]) unless MiGA::Project.exist? cli[:outdir]
     # Define project metadata
     p = cli.load_project(:outdir, '-o')
-    [:haai_p, :aai_p, :ani_p, :ess_coll].each { |i| p_metadata[i] = cli[i] }
+    %i[haai_p aai_p ani_p ess_coll min_qual].each { |i| p_metadata[i] = cli[i] }
     p_metadata[:type] = cli[:project_type]
     transfer_metadata(p, p_metadata)
     # Download datasets
@@ -159,7 +166,7 @@ module MiGA::Cli::Action::Wf
                  '-P', cli[:outdir],
                  '-r', r,
                  '-o', File.expand_path("#{r}.tsv", cli[:outdir]),
-                 '--tab'
+                 '--tab', '--ref', '--active'
                ])
     end
   end

data/lib/miga/daemon.rb

@@ -72,6 +72,7 @@ class MiGA::Daemon < MiGA::MiGA
     say '-----------------------------------'
     say 'MiGA:%s launched' % project.name
     say '-----------------------------------'
+    recalculate_status!
     load_status
     say 'Configuration options:'
     say @runopts.to_s
@@ -99,6 +100,7 @@ class MiGA::Daemon < MiGA::MiGA
   end
 
   def recalculate_status!
+    say 'Recalculating status for all datasets'
     project.each_dataset(&:recalculate_status)
   end
 
@@ -158,8 +160,8 @@ class MiGA::Daemon < MiGA::MiGA
   end
 
   ##
-  # Traverse datasets, and returns boolean indicating if at any datasets
-  # are incomplete
+  # Traverse datasets, and returns boolean indicating if at any reference
+  # datasets are incomplete
   def check_datasets
     l_say(2, 'Checking datasets')
     o = false
@@ -167,7 +169,7 @@ class MiGA::Daemon < MiGA::MiGA
       next unless ds.status == :incomplete
       next if ds.next_preprocessing(false).nil?
 
-      o = true
+      o = true if ds.ref?
       queue_job(:d, ds)
     end
     o

data/lib/miga/dataset.rb

@@ -97,7 +97,10 @@ class MiGA::Dataset < MiGA::MiGA
 
   ##
   # Inactivate a dataset. This halts automated processing by the daemon
-  def inactivate!
+  # 
+  # If given, the +reason+ string is saved as a metadata +:warn+ entry
+  def inactivate!(reason = nil)
+    metadata[:warn] = "Inactive: #{reason}" unless reason.nil?
     metadata[:inactive] = true
     metadata.save
     pull_hook :on_inactivate
@@ -107,6 +110,7 @@ class MiGA::Dataset < MiGA::MiGA
   # Activate a dataset. This removes the +:inactive+ flag
   def activate!
     metadata[:inactive] = nil
+    metadata[:warn] = nil if metadata[:warn] && metadata[:warn] =~ /^Inactive: /
     metadata.save
     pull_hook :on_activate
   end

data/lib/miga/dataset/base.rb

@@ -35,8 +35,8 @@ module MiGA::Dataset::Base
     mytaxa: '07.annotation/02.taxonomy/01.mytaxa',
     mytaxa_scan: '07.annotation/03.qa/02.mytaxa_scan',
     # Distances (for single-species datasets)
-    distances: '09.distances',
     taxonomy: '09.distances/05.taxonomy',
+    distances: '09.distances',
     # General statistics
     stats: '90.stats'
   }

data/lib/miga/dataset/hooks.rb

@@ -52,15 +52,15 @@ module MiGA::Dataset::Hooks
   end
 
   ##
-  # Run +cmd+ in the command-line with {{variables}}: dataset, project, miga,
-  # object (as defined for the event, if any)
+  # Run +cmd+ in the command-line with {{variables}}:
+  # dataset, project, project_name, miga, object (if defined for the event)
   # - +hook_args+: +[cmd]+
   # - +event_args+: +[object (optional)]+
   def hook_run_cmd(hook_args, event_args)
     Process.wait(
       spawn hook_args.first.miga_variables(
-        dataset: name, project: project.path, miga: MiGA::MiGA.root_path,
-        object: event_args.first
+        dataset: name, project: project.path, project_name: project.name,
+        miga: MiGA::MiGA.root_path, object: event_args.first
       )
     )
   end

data/lib/miga/dataset/result.rb

@@ -70,7 +70,7 @@ module MiGA::Dataset::Result
         false
       elsif add_result(t, save).nil?
         if (metadata["_try_#{t}"] || 0) > (project.metadata[:max_try] || 10)
-          inactivate!
+          inactivate! "Too many errors in step #{t}"
           false
         else
           true

data/lib/miga/project/hooks.rb

@@ -26,14 +26,15 @@ module MiGA::Project::Hooks
   end
 
   ##
-  # Run +cmd+ in the command-line with {{variables}}: project, miga,
-  # object (as defined by the event, if any)
+  # Run +cmd+ in the command-line with {{variables}}:
+  # project, project_name, miga, object (if defined by the event)
   # - +hook_args+: +[cmd]+
   # - +event_args+: +[object (optional)]+
   def hook_run_cmd(hook_args, event_args)
     Process.wait(
       spawn hook_args.first.miga_variables(
-        project: path, miga: MiGA::MiGA.root_path, object: event_args.first
+        project: path, project_name: name,
+        miga: MiGA::MiGA.root_path, object: event_args.first
       )
     )
   end

data/lib/miga/remote_dataset/download.rb

@@ -94,12 +94,13 @@ class MiGA::RemoteDataset
       @timeout_try = 0
       begin
         DEBUG 'GET: ' + url
-        open(url, read_timeout: 600) { |f| doc = f.read }
+        URI.parse(url).open(read_timeout: 600) { |f| doc = f.read }
       rescue => e
         @timeout_try += 1
         raise e if @timeout_try >= 3
 
         sleep 5 # <- For: 429 Too Many Requests
+        DEBUG "RETRYING after: #{e}"
         retry
       end
       doc

data/lib/miga/result.rb

@@ -164,7 +164,9 @@ class MiGA::Result < MiGA::MiGA
   # Unlink result by removing the .done and .start timestamps and the
   # .json descriptor, but don't remove any other associated files
   def unlink
-    %i(start done).each { |i| f = path(i) and File.unlink(f) }
+    %i(start done).each do |i|
+      f = path(i) and File.exists?(f) and File.unlink(f)
+    end
     File.unlink path
   end
 

data/lib/miga/result/stats.rb

@@ -8,6 +8,7 @@ module MiGA::Result::Stats
   # (Re-)calculate and save the statistics for the result
   def compute_stats
     method = :"compute_stats_#{key}"
+    MiGA::MiGA.DEBUG "Result(#{key}).compute_stats"
     stats = self.respond_to?(method, true) ? send(method) : nil
     unless stats.nil?
       self[:stats] = stats
@@ -109,20 +110,8 @@ module MiGA::Result::Stats
         end
       end
     else
-      # Fix estimate by domain
-      if !(tax = source.metadata[:tax]).nil? &&
-         %w[Archaea Bacteria].include?(tax[:d]) &&
-         file_path(:raw_report).nil?
-        scr = "#{MiGA::MiGA.root_path}/utils/domain-ess-genes.rb"
-        rep = file_path(:report)
-        rc_p = File.expand_path('.miga_rc', ENV['HOME'])
-        rc = File.exist?(rc_p) ? ". '#{rc_p}' && " : ''
-        $stderr.print `#{rc} ruby '#{scr}' \
-          '#{rep}' '#{rep}.domain' '#{tax[:d][0]}'`
-        add_file(:raw_report, "#{source.name}.ess/log")
-        add_file(:report, "#{source.name}.ess/log.domain")
-      end
-      # Extract/compute quality values
+      # Estimate quality metrics
+      fix_essential_genes_by_domain
       stats = { completeness: [0.0, '%'], contamination: [0.0, '%'] }
       File.open(file_path(:report), 'r') do |fh|
         fh.each_line do |ln|
@@ -131,6 +120,8 @@ module MiGA::Result::Stats
           end
         end
       end
+
+      # Determine qualitative range
       stats[:quality] = stats[:completeness][0] - stats[:contamination][0] * 5
       source.metadata[:quality] =
         case stats[:quality]
@@ -140,6 +131,12 @@ module MiGA::Result::Stats
         else; :low
         end
       source.save
+
+      # Inactivate low-quality datasets
+      min_qual = (project.metadata[:min_qual] || 50)
+      if min_qual != 'no' && stats[:quality] < min_qual
+        source.inactivate! 'Low genome quality'
+      end
     end
     stats
   end
@@ -175,4 +172,21 @@ module MiGA::Result::Stats
     end
     stats
   end
+
+  # Fix estimates based on essential genes based on taxonomy
+  def fix_essential_genes_by_domain
+    return if (tax = source.metadata[:tax]).nil? ||
+       !%w[Archaea Bacteria].include?(tax[:d]) ||
+       file_path(:raw_report)
+
+    MiGA::MiGA.DEBUG "Fixing essential genes by domain"
+    scr = "#{MiGA::MiGA.root_path}/utils/domain-ess-genes.rb"
+    rep = file_path(:report)
+    rc_p = File.expand_path('.miga_rc', ENV['HOME'])
+    rc = File.exist?(rc_p) ? ". '#{rc_p}' && " : ''
+    $stderr.print `#{rc} ruby '#{scr}' \
+      '#{rep}' '#{rep}.domain' '#{tax[:d][0]}'`
+    add_file(:raw_report, "#{source.name}.ess/log")
+    add_file(:report, "#{source.name}.ess/log.domain")
+  end
 end

data/lib/miga/version.rb

@@ -8,7 +8,7 @@ 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, 9, 0]
+  VERSION = [0.7, 10, 0]
 
   ##
   # Nickname for the current major.minor version.
@@ -16,7 +16,7 @@ module MiGA
 
   ##
   # Date of the current gem release.
-  VERSION_DATE = Date.new(2020, 6, 8)
+  VERSION_DATE = Date.new(2020, 6, 29)
 
   ##
   # Reference of MiGA.

data/scripts/distances.bash

@@ -9,7 +9,12 @@ cd "$PROJECT/data/09.distances"
 # Initialize
 miga date > "$DATASET.start"
 
-# Run
+# Check quality first
+miga stats -P "$PROJECT" -D "$DATASET" -r essential_genes --compute-and-save
+inactive=$(miga ls -P "$PROJECT" -D "$DATASET" -m inactive | cut -f 2)
+[[ "$inactive" == "true" ]] && exit
+
+# Run distances
 ruby -I "$MIGA/lib" "$MIGA/utils/distances.rb" "$PROJECT" "$DATASET"
 
 # Finalize

data/test/daemon_test.rb

@@ -93,7 +93,7 @@ class DaemonTest < Test::Unit::TestCase
       0 => /-{20}\n/,
       1 => /MiGA:#{p.name} launched/,
       2 => /-{20}\n/,
-      5 => /Probing running jobs\n/
+      6 => /Probing running jobs\n/
     }.each { |k, v| assert_match(v, l[k], "unexpected line: #{k}") }
   ensure
     begin

data/test/dataset_test.rb

@@ -185,11 +185,13 @@ class DatasetTest < Test::Unit::TestCase
     d = dataset
     assert_equal(:incomplete, d.status)
     assert_predicate(d, :active?)
-    d.inactivate!
+    d.inactivate! 'Too annoying'
     assert_equal(:inactive, d.status)
+    assert_equal('Inactive: Too annoying', d.metadata[:warn])
     assert_not_predicate(d, :active?)
     d.activate!
     assert_equal(:incomplete, d.status)
+    assert_nil(d.metadata[:warn])
     assert_predicate(d, :active?)
   end
 

data/test/remote_dataset_test.rb

@@ -101,7 +101,7 @@ class RemoteDatasetTest < Test::Unit::TestCase
 
   def test_ref_type_status
     declare_remote_access
-    rd = MiGA::RemoteDataset.new('GCA_002849345', :assembly, :ncbi)
+    rd = MiGA::RemoteDataset.new('GCA_003144295.1', :assembly, :ncbi)
     assert { !rd.get_metadata[:is_type] }
     assert { rd.get_metadata[:is_ref_type] }
   end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: miga-base
 version: !ruby/object:Gem::Version
-  version: 0.7.9.0
+  version: 0.7.10.0
 platform: ruby
 authors:
 - Luis M. Rodriguez-R
-autorequire: 
+autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-06-08 00:00:00.000000000 Z
+date: 2020-06-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: daemons
@@ -529,7 +529,7 @@ homepage: http://enve-omics.ce.gatech.edu/miga
 licenses:
 - Artistic-2.0
 metadata: {}
-post_install_message: 
+post_install_message:
 rdoc_options:
 - lib
 - README.md
@@ -550,8 +550,8 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.0.3
-signing_key: 
+rubygems_version: 3.1.2
+signing_key:
 specification_version: 4
 summary: MiGA
 test_files: []