miga-base 0.7.9.0 → 0.7.12.1

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
@@ -16,32 +17,45 @@ module MiGA::Result::Stats
     self[:stats]
   end
 
+  ##
+  # Access the stats entry of results
+  def stats
+    self[:stats]
+  end
+
   private
 
   def compute_stats_raw_reads
     stats = {}
+    seq_opts = { gc: true, x: true, skew: true }
     if self[:files][:pair1].nil?
-      s = MiGA::MiGA.seqs_length(file_path(:single), :fastq, gc: true, x: true)
+      s = MiGA::MiGA.seqs_length(file_path(:single), :fastq, seq_opts)
       stats = {
         reads: s[:n],
         length_average: [s[:avg], 'bp'],
         length_standard_deviation: [s[:sd], 'bp'],
         g_c_content: [s[:gc], '%'],
-        x_content: [s[:x], '%']
+        x_content: [s[:x], '%'],
+        g_c_skew: [s[:gc_skew], '%'],
+        a_t_skew: [s[:at_skew], '%']
       }
     else
-      s1 = MiGA::MiGA.seqs_length(file_path(:pair1), :fastq, gc: true, x: true)
-      s2 = MiGA::MiGA.seqs_length(file_path(:pair2), :fastq, gc: true, x: true)
+      s1 = MiGA::MiGA.seqs_length(file_path(:pair1), :fastq, seq_opts)
+      s2 = MiGA::MiGA.seqs_length(file_path(:pair2), :fastq, seq_opts)
       stats = {
         read_pairs: s1[:n],
         forward_length_average: [s1[:avg], 'bp'],
         forward_length_standard_deviation: [s1[:sd], 'bp'],
         forward_g_c_content: [s1[:gc], '%'],
         forward_x_content: [s1[:x], '%'],
+        forward_g_c_skew: [s1[:gc_skew], '%'],
+        forward_a_t_skew: [s1[:at_skew], '%'],
         reverse_length_average: [s2[:avg], 'bp'],
         reverse_length_standard_deviation: [s2[:sd], 'bp'],
         reverse_g_c_content: [s2[:gc], '%'],
-        reverse_x_content: [s2[:x], '%']
+        reverse_x_content: [s2[:x], '%'],
+        reverse_g_c_skew: [s2[:gc_skew], '%'],
+        reverse_a_t_skew: [s2[:at_skew], '%']
       }
     end
     stats
@@ -49,19 +63,22 @@ module MiGA::Result::Stats
 
   def compute_stats_trimmed_fasta
     f = self[:files][:coupled].nil? ? file_path(:single) : file_path(:coupled)
-    s = MiGA::MiGA.seqs_length(f, :fasta, gc: true, x: true)
+    s = MiGA::MiGA.seqs_length(f, :fasta, gc: true, x: true, skew: true)
     {
       reads: s[:n],
       length_average: [s[:avg], 'bp'],
       length_standard_deviation: [s[:sd], 'bp'],
       g_c_content: [s[:gc], '%'],
-      x_content: [s[:x], '%']
+      x_content: [s[:x], '%'],
+      g_c_skew: [s[:gc_skew], '%'],
+      a_t_skew: [s[:at_skew], '%']
     }
   end
 
   def compute_stats_assembly
     s = MiGA::MiGA.seqs_length(
-      file_path(:largecontigs), :fasta, n50: true, gc: true, x: true
+      file_path(:largecontigs), :fasta,
+      n50: true, gc: true, x: true, skew: true
     )
     {
       contigs: s[:n],
@@ -69,7 +86,9 @@ module MiGA::Result::Stats
       total_length: [s[:tot], 'bp'],
       longest_sequence: [s[:max], 'bp'],
       g_c_content: [s[:gc], '%'],
-      x_content: [s[:x], '%']
+      x_content: [s[:x], '%'],
+      g_c_skew: [s[:gc_skew], '%'],
+      a_t_skew: [s[:at_skew], '%']
     }
   end
 
@@ -109,20 +128,8 @@ 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 +138,8 @@ 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 +149,12 @@ module MiGA::Result::Stats
         else; :low
         end
       source.save
+
+      # Inactivate low-quality datasets
+      min_qual = (project.metadata[:min_qual] || 25)
+      if min_qual != 'no' && stats[:quality] < min_qual
+        source.inactivate! 'Low quality genome'
+      end
     end
     stats
   end
@@ -175,4 +190,21 @@ 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, 12, 1]
 
   ##
   # 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, 7, 24)
 
   ##
   # Reference of MiGA.

data/scripts/cds.bash

@@ -20,7 +20,6 @@ fi
 TYPE=$(miga ls -P "$PROJECT" -D "$DATASET" -m type | cut -f 2)
 case "$TYPE" in
   metagenome|virome)
-    $CMD -p meta
     prodigal -a "${DATASET}.faa" -d "${DATASET}.fna" -o "${DATASET}.gff3" \
       -f gff -q -i "../05.assembly/${DATASET}.LargeContigs.fna" -p meta
     ;;

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/project_test.rb

@@ -108,7 +108,7 @@ class ProjectTest < Test::Unit::TestCase
     d1 = p1.add_dataset('BAH')
     assert_not_predicate(p1, :done_preprocessing?)
     FileUtils.touch(File.join(p1.path, 'data', '90.stats', "#{d1.name}.done"))
-    assert_predicate(p1, :done_preprocessing?)
+    assert { p1.done_preprocessing? true }
     assert_nil(p1.next_inclade)
     p1.metadata[:type] = :clade
     assert_equal(:subclades, p1.next_inclade)

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: 0.7.12.1
 platform: ruby
 authors:
 - Luis M. Rodriguez-R
-autorequire: 
+autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-06-08 00:00:00.000000000 Z
+date: 2020-07-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: daemons
@@ -118,6 +118,16 @@ files:
 - lib/miga/cli/action/add.rb
 - lib/miga/cli/action/add_result.rb
 - lib/miga/cli/action/archive.rb
+- lib/miga/cli/action/browse.rb
+- lib/miga/cli/action/browse/about.html
+- lib/miga/cli/action/browse/dataset.html
+- lib/miga/cli/action/browse/dataset_menu_item.html
+- lib/miga/cli/action/browse/datasets.html
+- lib/miga/cli/action/browse/favicon-32.png
+- lib/miga/cli/action/browse/index.html
+- lib/miga/cli/action/browse/layout.html
+- lib/miga/cli/action/browse/redirect.html
+- lib/miga/cli/action/browse/style.css
 - lib/miga/cli/action/classify_wf.rb
 - lib/miga/cli/action/console.rb
 - lib/miga/cli/action/daemon.rb
@@ -529,7 +539,7 @@ 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 +560,8 @@ 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: []