miga-base 0.7.3.1 → 0.7.8.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 7e3f943800139b48778d227b3f0680bd71cac91774037f9f9e115842ccb3e6f4
-  data.tar.gz: 351042f7e77d3b2a3189a27fc37ead6c4fdb02ad519ba72a760d8e75a50dddf3
+  metadata.gz: 5cce1edf29f5f41ad7a53225978513597f2234abd781c5f179c9f45f9d6f1ec2
+  data.tar.gz: 846a570b68efe9dba8acbc1434c55752373d9864df2cea2ff467330bf7e8b1ac
 SHA512:
-  metadata.gz: 95f0728ad6f0d4359c32ed37deb9b215e6ecbd15fd5807796a8927eadb5ebfc1466d88e94afec179b33df9009d53e2f109166c5db5bf82424d6c31c6289cf780
-  data.tar.gz: 75a5c9f95150c0177f1cf29a2bac58547bb51e60f75fe1d3c42acc3a3945aee161d51e1af479187459312b39c9164a6955d28232cbeb923cb8dbe5b748b590d8
+  metadata.gz: 123117d43c4ef4200ebe4a51cb14d19e48b89d249e66fa27b6e0834cf1c3b738535c3f514856e373259428c42543a25886744b4a8a6775f75b587d0a557e4d48
+  data.tar.gz: fa178d1e899da25212983bf7640cc391f3922e560d5d5ee4ff072781b6301013d928212868c9db1590dfcd1f41449e07b5019765dc866d02295226c9ed8a3e2d

data/lib/miga/cli.rb

@@ -7,7 +7,6 @@ require 'optparse'
 ##
 # MiGA Command Line Interface API.
 class MiGA::Cli < MiGA::MiGA
-
   require 'miga/cli/base'
   require 'miga/cli/opt_helper'
   require 'miga/cli/objects_helper'
@@ -61,7 +60,7 @@ class MiGA::Cli < MiGA::MiGA
 
   def initialize(argv)
     @data = {}
-    @defaults = {verbose: false, tabular: false}
+    @defaults = { verbose: false, tabular: false }
     @opt_common = true
     @objects = {}
     if argv[0].nil? or argv[0].to_s[0] == '-'
@@ -106,6 +105,7 @@ class MiGA::Cli < MiGA::MiGA
   # otherwise it's sent to +$stderr+
   def say(*par)
     return unless self[:verbose]
+
     super(*par)
   end
 
@@ -116,16 +116,17 @@ class MiGA::Cli < MiGA::MiGA
   # The report goes to $stderr iff --verborse
   def advance(step, n = 0, total = nil, bin = true)
     return unless self[:verbose]
+
     adv = total.nil? ? (n == 0 ? '' : num_suffix(n, bin)) :
       ('%.1f%% (%s/%s)' % [100.0 * n / total,
-        num_suffix(n, bin), num_suffix(total, bin)])
+                           num_suffix(n, bin), num_suffix(total, bin)])
     $stderr.print("[%s] %s %s    \r" % [Time.now, step, adv])
   end
 
   def num_suffix(n, bin = false)
     p = ''
-    {T: 4, G: 3, M: 2, K: 1}.each do |k,x|
-      v = (bin ? 1024 : 1e3) ** x
+    { T: 4, G: 3, M: 2, K: 1 }.each do |k, x|
+      v = (bin ? 1024 : 1e3)**x
       if n > v
         n = '%.1f' % (n / v)
         p = k
@@ -160,7 +161,7 @@ class MiGA::Cli < MiGA::MiGA
   ##
   # Set default values in the Hash +hsh+
   def defaults=(hsh)
-    hsh.each{ |k,v| @defaults[k] = v }
+    hsh.each { |k, v| @defaults[k] = v }
   end
 
   ##
@@ -192,6 +193,7 @@ class MiGA::Cli < MiGA::MiGA
   def launch
     begin
       raise "See `miga -h`" if action.nil?
+
       action.launch
     rescue => err
       $stderr.puts "Exception: #{err}"
@@ -222,8 +224,8 @@ class MiGA::Cli < MiGA::MiGA
   # +par+, a Hash with object names as keys and parameter flag as values.
   # If missing, raise an error with message +msg+
   def ensure_par(req, msg = '%<name>s is mandatory: please provide %<flag>s')
-    req.each do |k,v|
-      raise (msg % {name: k, flag: v}) if self[k].nil?
+    req.each do |k, v|
+      raise (msg % { name: k, flag: v }) if self[k].nil?
     end
   end
 

data/lib/miga/cli/action.rb

@@ -8,7 +8,6 @@ require 'miga/cli'
 # by MiGA::Cli::Action::* classes. Do not attempt creating directly with +new+,
 # use instead the MiGA::Cli::Action.load interface.
 class MiGA::Cli::Action < MiGA::MiGA
-
   class << self
     def load(task, cli)
       require "miga/cli/action/#{task}"
@@ -55,8 +54,8 @@ class MiGA::Cli::Action < MiGA::MiGA
   ##
   # Name of the action, as referred to by the CLI
   def name
-    camel = self.class.to_s.gsub(/.*::/,'')
-    camel.gsub(/(\S)([A-Z])/,'\1_\2').downcase
+    camel = self.class.to_s.gsub(/.*::/, '')
+    camel.gsub(/(\S)([A-Z])/, '\1_\2').downcase
   end
 
   ##

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

@@ -4,23 +4,22 @@
 require 'miga/cli/action'
 
 class MiGA::Cli::Action::About < MiGA::Cli::Action
-
   def parse_cli
-    cli.defaults = {info: false, processing: false, tabular: false}
+    cli.defaults = { info: false, processing: false, tabular: false }
     cli.parse do |opt|
       cli.opt_object(opt, [:project])
       opt.on(
         '-p', '--processing',
         'Print information on processing advance'
-        ){ |v| cli[:processing] = v }
+      ) { |v| cli[:processing] = v }
       opt.on(
         '-m', '--metadata STRING',
         'Print name and metadata field only'
-        ){ |v| cli[:datum] = v }
+      ) { |v| cli[:datum] = v }
       opt.on(
         '--tab',
         'Return a tab-delimited table'
-        ){ |v| cli[:tabular] = v }
+      ) { |v| cli[:tabular] = v }
     end
   end
 
@@ -37,7 +36,7 @@ class MiGA::Cli::Action::About < MiGA::Cli::Action
     else
       cli.puts MiGA.tabulate([:key, :value], p.metadata.data.keys.map do |k|
         v = p.metadata[k]
-        [k, k==:datasets ? v.size : v]
+        [k, k == :datasets ? v.size : v]
       end, cli[:tabular])
     end
   end

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

@@ -4,7 +4,6 @@
 require 'miga/cli/action'
 
 class MiGA::Cli::Action::Add < MiGA::Cli::Action
-
   def parse_cli
     cli.expect_files = true
     cli.defaults = {
@@ -47,7 +46,7 @@ class MiGA::Cli::Action::Add < MiGA::Cli::Action
       opt.on(
         '-i', '--input-type STRING',
         'Type of input data, one of the following:',
-        *self.class.INPUT_TYPES.map{ |k,v| "~ #{k}: #{v[0]}" }
+        *self.class.INPUT_TYPES.map { |k, v| "~ #{k}: #{v[0]}" }
       ) { |v| cli[:input_type] = v.downcase.to_sym }
       opt.on(
         '--ignore-dups',
@@ -64,6 +63,7 @@ class MiGA::Cli::Action::Add < MiGA::Cli::Action
     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
@@ -76,22 +76,22 @@ class MiGA::Cli::Action::Add < MiGA::Cli::Action
   @@INPUT_TYPES = {
     raw_reads_single:
       ['Single raw reads in a single FastQ file',
-        :raw_reads, %w[.1.fastq]],
+       :raw_reads, %w[.1.fastq]],
     raw_reads_paired:
       ['Paired raw reads in two FastQ files',
-        :raw_reads, %w[.1.fastq .2.fastq]],
+       :raw_reads, %w[.1.fastq .2.fastq]],
     trimmed_reads_single:
       ['Single trimmed reads in a single FastA file',
-        :trimmed_fasta, %w[.SingleReads.fa]],
+       :trimmed_fasta, %w[.SingleReads.fa]],
     trimmed_reads_paired:
       ['Paired trimmed reads in two FastA files',
-        :trimmed_fasta, %w[.1.fasta .2.fasta]],
+       :trimmed_fasta, %w[.1.fasta .2.fasta]],
     trimmed_reads_interleaved:
       ['Paired trimmed reads in a single FastA file',
-        :trimmed_fasta, %w[.CoupledReads.fa]],
+       :trimmed_fasta, %w[.CoupledReads.fa]],
     assembly:
       ['Assembled contigs or scaffolds in FastA format',
-        :assembly, %w[.LargeContigs.fna]]
+       :assembly, %w[.LargeContigs.fna]]
   }
 
   class << self
@@ -106,23 +106,26 @@ class MiGA::Cli::Action::Add < MiGA::Cli::Action
     files = cli.files
     file_type = nil
     if files.empty?
-      cli.ensure_par({dataset: '-D'},
-        'dataset is mandatory (-D) unless files are provided')
+      cli.ensure_par({ dataset: '-D' },
+                     'dataset is mandatory (-D) unless files are provided')
       cli.ensure_type(Dataset)
       files = [nil]
     else
       raise 'Please specify input type (-i).' if cli[:input_type].nil?
+
       file_type = self.class.INPUT_TYPES[cli[:input_type]]
       raise "Unrecognized input type: #{cli[:input_type]}." if file_type.nil?
       raise 'Some files are duplicated, files must be unique.' if
         files.size != files.uniq.size
+
       if cli[:input_type].to_s =~ /_paired$/
         if files.size.odd?
           raise 'Odd number of files incompatible with input type.'
         end
+
         files = Hash[*files].to_a
       else
-        files = files.map{ |i| [i] }
+        files = files.map { |i| [i] }
       end
       if files.size > 1 && !cli[:dataset].nil?
         raise 'The dataset name (-D) can only be specified with one input file.'
@@ -137,6 +140,7 @@ class MiGA::Cli::Action::Add < MiGA::Cli::Action
       ref_file = file.is_a?(Array) ? file.first : file
       m = cli[:regexp].match(ref_file)
       raise "Cannot extract name from file: #{ref_file}" if m.nil? or m[1].nil?
+
       name = cli[:prefix].to_s + m[1].miga_name
     end
     if Dataset.exist?(p, name)
@@ -151,12 +155,14 @@ class MiGA::Cli::Action::Add < MiGA::Cli::Action
     cli.say "o #{name}"
     d = Dataset.new(p, name, cli[:ref])
     raise "Unexpected: Couldn't create dataset: #{name}." if d.nil?
+
     d
   end
 
   def copy_file_to_project(file, file_type, d, p)
     return if file.nil?
-    r_dir = Dataset.RESULT_DIRS[ file_type[1] ]
+
+    r_dir = Dataset.RESULT_DIRS[file_type[1]]
     r_path = File.expand_path("data/#{r_dir}/#{d.name}", p.path)
     file_type[2].each_with_index do |ext, i|
       gz = file[i] =~ /\.gz/ ? '.gz' : ''

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

@@ -4,15 +4,14 @@
 require 'miga/cli/action'
 
 class MiGA::Cli::Action::AddResult < MiGA::Cli::Action
-
   def parse_cli
-    cli.defaults = {force: false}
+    cli.defaults = { force: false }
     cli.parse do |opt|
       cli.opt_object(opt, [:project, :dataset_opt, :result])
       opt.on(
         '-f', '--force',
         'Force re-indexing of the result even if it\'s already registered'
-        ){ |v| cli[:force] = v }
+      ) { |v| cli[:force] = v }
     end
   end
 

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

@@ -5,7 +5,6 @@ require 'miga/cli/action'
 require 'rubygems/package'
 
 class MiGA::Cli::Action::Archive < MiGA::Cli::Action
-
   def parse_cli
     cli.parse do |opt|
       opt.on(
@@ -26,6 +25,7 @@ class MiGA::Cli::Action::Archive < MiGA::Cli::Action
     unless cli[:tarball] =~ /\.tar\.gz$/
       raise 'The tarball path (-o) must have .tar.gz extension'
     end
+
     cli[:folder] ||= cli.load_project.name
     ds = cli.load_and_filter_datasets
 
@@ -95,5 +95,4 @@ class MiGA::Cli::Action::Archive < MiGA::Cli::Action
     in_tar = File.join(cli[:folder], rel_path)
     tar.add_file_simple(in_tar, 0666, string.size) { |fh| fh.write(string) }
   end
-
 end

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

@@ -51,24 +51,24 @@ class MiGA::Cli::Action::ClassifyWf < MiGA::Cli::Action
     ref_db = reference_db
     p_metadata = Hash[
       %w[project_stats haai_distances aai_distances ani_distances clade_finding]
-        .map { |i| ["run_#{i}", false] }
+                 .map { |i| ["run_#{i}", false] }
     ]
     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)
+                       run_ssu: false, run_mytaxa_scan: false, run_distances: false)
     # 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])
-    ])
+               'ls', '-P', cli[:outdir], '-m', 'tax', '--tab',
+               '-o', File.expand_path('classification.tsv', cli[:outdir])
+             ])
     cleanup
   end
-  
+
   private
 
   def reference_db
@@ -85,12 +85,14 @@ class MiGA::Cli::Action::ClassifyWf < MiGA::Cli::Action
         unless File.size? lm_f
           raise 'No locally listed databases, call "miga get_db" first'
         end
+
         cli[:database] = MiGA::Json.parse(lm_f)[:databases].keys.first
       end
       ref_db_path = File.expand_path(cli[:database].to_s, cli[:local])
     end
     ref_db = MiGA::Project.load(ref_db_path)
     raise "Cannot locate reference database: #{ref_db_path}" if ref_db.nil?
+
     cli.say "Reference database: #{ref_db.name}"
     ref_db
   end

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

@@ -4,7 +4,6 @@
 require 'miga/cli/action'
 
 class MiGA::Cli::Action::Console < MiGA::Cli::Action
-
   def parse_cli
   end
 

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

@@ -5,17 +5,17 @@ require 'miga/cli/action'
 require 'miga/daemon'
 
 class MiGA::Cli::Action::Daemon < MiGA::Cli::Action
-
   def parse_cli
-    cli.defaults = {daemon_opts: []}
+    cli.defaults = { daemon_opts: [] }
     cli.expect_operation = true
     cli.parse do |opt|
       opt.separator 'Available operations:'
-      { start:   'Start an instance of the application',
-        stop:    'Start an instance of the application',
-        run:     'Start the application and stay on top',
-        status:  'Show status (PID) of application instances'
-      }.each { |k,v| opt.separator sprintf '    %*s%s', -33, k, v }
+      {
+        start: 'Start an instance of the application',
+        stop: 'Start an instance of the application',
+        run: 'Start the application and stay on top',
+        status: 'Show status (PID) of application instances'
+      }.each { |k, v| opt.separator sprintf('    %*s%s', -33, k, v) }
       opt.separator ''
 
       opt.separator 'MiGA options:'

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

@@ -4,7 +4,6 @@
 require 'miga/cli/action'
 
 class MiGA::Cli::Action::Date < MiGA::Cli::Action
-
   def parse_cli
     cli.parse { |_| }
   end

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

@@ -44,12 +44,13 @@ class MiGA::Cli::Action::DerepWf < MiGA::Cli::Action
   def perform
     # Input data
     p = create_project(:assembly,
-      { run_project_stats: false, run_clades: false,
-        gsp_metric: cli[:metric], :"gsp_#{cli[:metric]}" => cli[:threshold] },
-      { run_mytaxa_scan: false, run_ssu: false })
+                       { run_project_stats: false, run_clades: false,
+                         gsp_metric: cli[:metric], :"gsp_#{cli[:metric]}" => cli[:threshold] },
+                       { 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
+
     # Run
     run_daemon
     dereplicate(p)
@@ -68,7 +69,7 @@ class MiGA::Cli::Action::DerepWf < MiGA::Cli::Action
     File.open(File.expand_path('genomospecies.tsv', cli[:outdir]), 'w') do |fh|
       fh.puts "Clade\tRepresentative\tMembers"
       clades.each_with_index do |i, k|
-        fh.puts ["gsp_#{k+1}", rep[k], i.join(',')].join("\t")
+        fh.puts ["gsp_#{k + 1}", rep[k], i.join(',')].join("\t")
       end
     end
     if cli[:collection]

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

@@ -1,15 +1,15 @@
 # @package MiGA
 # @license Artistic-2.0
 
-require 'miga/cli/action'
-require 'sqlite3'
+require 'miga/cli/action/doctor/base'
 
 class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
+  include MiGA::Cli::Action::Doctor::Base
 
   def parse_cli
-    @@OPERATIONS.keys.each { |i| cli.defaults = { i => true } }
+    cli.defaults = Hash[@@OPERATIONS.keys.map { |i| [i, true] }]
     cli.parse do |opt|
-      operation_n = Hash[@@OPERATIONS.map { |k,v| [v[0], k] }]
+      operation_n = Hash[@@OPERATIONS.map { |k, v| [v[0], k] }]
       cli.opt_object(opt, [:project])
       opt.on(
         '--ignore TASK1,TASK2', Array,
@@ -21,20 +21,12 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
         'Perform only the specified task (see --ignore)'
       ) do |v|
         op_k = @@OPERATIONS.find { |_, i| i[0] == v.downcase }.first
-        @@OPERATIONS.keys.each { |i| cli[i] = false }
+        @@OPERATIONS.each_key { |i| cli[i] = false }
         cli[op_k] = true
       end
     end
   end
 
-  def check_sqlite3_database(db_file, metric)
-    SQLite3::Database.new(db_file) do |conn|
-      conn.execute("select count(*) from #{metric}").first
-    end
-  rescue SQLite3::SQLException
-    yield
-  end
-
   def perform
     p = cli.load_project
     @@OPERATIONS.keys.each do |k|
@@ -43,7 +35,8 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
   end
 
   @@OPERATIONS = {
-    db: ['databases', 'Check database files integrity'],
+    status: ['status', 'Update metadata status of all datasets'],
+    db: ['databases', 'Check integrity of database files'],
     dist: ['distances', 'Check distance summary tables'],
     files: ['files', 'Check for outdated files'],
     cds: ['cds', 'Check for gzipped genes and proteins'],
@@ -52,36 +45,54 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
     start: ['start', 'Check for lingering .start files'],
     tax: ['taxonomy', 'Check for taxonomy consistency (not yet implemented)']
   }
+
   class << self
+    ##
+    # All supported operations
     def OPERATIONS
       @@OPERATIONS
     end
   end
 
+  ##
+  # Perform status operation with MiGA::Cli +cli+
+  def check_status(cli)
+    cli.say 'Updating metadata status'
+    n, k = cli.load_project.dataset_names.size, 0
+    cli.load_project.each_dataset do |d|
+      cli.advance('Datasets:', k += 1, n, false)
+      d.recalculate_status
+    end
+    cli.say
+  end
+
+  ##
+  # Perform databases operation with MiGA::Cli +cli+
   def check_db(cli)
-    cli.say 'Checking databases integrity'
+    cli.say 'Checking integrity of databases'
+    n, k = cli.load_project.dataset_names.size, 0
     cli.load_project.each_dataset do |d|
-      [:distances, :taxonomy].each do |r_key|
-        r = d.result(r_key) or next
-        {haai_db: :aai, aai_db: :aai, ani_db: :ani}.each do |db_key, metric|
-          db_file = r.file_path(db_key) or next
-          check_sqlite3_database(db_file, metric) do
-            cli.say(
-              "  > Removing #{db_key} #{r_key} table for #{d.name}")
-            [db_file, r.path(:done), r.path].each do |f|
-              File.unlink(f) if File.exist? f
-            end # each |f|
-          end # check_sqlite3_database
-        end # each |db_key, metric|
-      end # each |r_key|
-    end # each |d|
+      cli.advance('Datasets:', k += 1, n, false)
+      each_database_file(d) do |db_file, metric, result|
+        check_sqlite3_database(db_file, metric) do
+          cli.say("  > Removing malformed database from #{d.name}:#{result}   ")
+          File.unlink(db_file)
+          r = d.result(result) or next
+          [r.path(:done), r.path].each { |f| File.unlink(f) if File.exist?(f) }
+        end
+      end
+    end
+    cli.say
   end
 
+  ##
+  # Perform distances operation with MiGA::Cli +cli+
   def check_dist(cli)
     p = cli.load_project
-    [:ani, :aai].each do |dist|
+    %i[ani aai].each do |dist|
       res = p.result("#{dist}_distances")
       next if res.nil?
+
       cli.say "Checking #{dist} table for consistent datasets"
       notok, fix = check_dist_eval(cli, p, res)
       check_dist_fix(cli, p, fix)
@@ -89,10 +100,13 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
     end
   end
 
+  ##
+  # Perform files operation with MiGA::Cli +cli+
   def check_files(cli)
     cli.say 'Looking for outdated files in results'
-    p = cli.load_project
-    p.each_dataset do |d|
+    n, k = cli.load_project.dataset_names.size, 0
+    cli.load_project.each_dataset do |d|
+      cli.advance('Datasets:', k += 1, n, false)
       d.each_result do |r_k, r|
         ok = true
         r.each_file do |_f_sym, _f_rel, f_abs|
@@ -102,23 +116,28 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
           end
         end
         unless ok
-          cli.say "  > Registering again #{d.name}:#{r_k}"
+          cli.say "  > Registering again #{d.name}:#{r_k}   "
           d.add_result(r_k, true, force: true)
           sr = d.result(:stats) and sr.remove!
         end
       end
     end
+    cli.say
   end
 
+  ##
+  # Perform cds operation with MiGA::Cli +cli+
   def check_cds(cli)
     cli.say 'Looking for unzipped genes or proteins'
+    n, k = cli.load_project.dataset_names.size, 0
     cli.load_project.each_dataset do |d|
+      cli.advance('Datasets:', k += 1, n, false)
       res = d.result(:cds) or next
       changed = false
-      [:genes, :proteins, :gff3, :gff2, :tab].each do |f|
+      %i[genes proteins gff3 gff2 tab].each do |f|
         file = res.file_path(f) or next
         if file !~ /\.gz/
-          cli.say "  > Gzipping #{d.name} #{f}"
+          cli.say "  > Gzipping #{d.name} #{f}   "
           cmdo = `gzip -9 '#{file}'`.chomp
           warn(cmdo) unless cmdo.empty?
           changed = true
@@ -129,13 +148,17 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
         sr = d.result(:stats) and sr.remove!
       end
     end
+    cli.say
   end
 
+  ##
+  # Perform essential-genes operation with MiGA::Cli +cli+
   def check_ess(cli)
     cli.say 'Looking for unarchived essential genes'
     cli.load_project.each_dataset do |d|
       res = d.result(:essential_genes)
       next if res.nil?
+
       dir = res.file_path(:collection)
       if dir.nil?
         cli.say "  > Removing #{d.name}:essential_genes"
@@ -144,17 +167,21 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
         next
       end
       next if Dir["#{dir}/*.faa"].empty?
+
       cli.say "  > Fixing #{d.name}"
       cmdo = `cd '#{dir}' && tar -zcf proteins.tar.gz *.faa && rm *.faa`.chomp
       warn(cmdo) unless cmdo.empty?
     end
   end
 
+  ##
+  # Perform mytaxa-scan operation with MiGA::Cli +cli+
   def check_mts(cli)
     cli.say 'Looking for unarchived MyTaxa Scan runs'
     cli.load_project.each_dataset do |d|
       res = d.result(:mytaxa_scan)
       next if res.nil?
+
       dir = res.file_path(:regions)
       fix = false
       unless dir.nil?
@@ -166,8 +193,8 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
         end
         fix = true
       end
-      %w[blast mytaxain wintax gene_ids region_ids].each do |ext|
-        file = res.file_path(ext.to_sym)
+      %i[blast mytaxain wintax gene_ids region_ids].each do |ext|
+        file = res.file_path(ext)
         unless file.nil?
           FileUtils.rm(file) if File.exist? file
           fix = true
@@ -180,6 +207,8 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
     end
   end
 
+  ##
+  # Perform start operation with MiGA::Cli +cli+
   def check_start(cli)
     cli.say 'Looking for legacy .start files lingering'
     cli.load_project.each_dataset do |d|
@@ -192,52 +221,12 @@ class MiGA::Cli::Action::Doctor < MiGA::Cli::Action
     end
   end
 
+  ##
+  # Perform taxonomy operation with MiGA::Cli +cli+
   def check_tax(cli)
-    #cli.say 'o Checking for taxonomy/distances consistency'
+    # cli.say 'o Checking for taxonomy/distances consistency'
     # TODO: Find 95%ANI clusters with entries from different species
-  end
-
-  private
-
-  def check_dist_eval(cli, p, res)
-    notok = {}
-    fix = {}
-    Zlib::GzipReader.open(res.file_path(:matrix)) do |fh|
-      lineno = 0
-      fh.each_line do |ln|
-        next if (lineno += 1) == 1
-        r = ln.split("\t")
-        next unless [1, 2].map { |i| p.dataset(r[i]).nil? }.any?
-        [1, 2].each do |i|
-          if p.dataset(r[i]).nil?
-            notok[r[i]] = true
-          else
-            fix[r[i]] = true
-          end
-        end
-      end
-    end
-    [notok, fix]
-  end
-
-  def check_dist_fix(cli, p, fix)
-    return if fix.empty?
-    cli.say("- Fixing #{fix.size} datasets")
-    fix.keys.each do |d_n|
-      cli.say "  > Fixing #{d_n}."
-      p.dataset(d_n).cleanup_distances!
-    end
-  end
-
-  def check_dist_recompute(cli, p, notok)
-    return if notok.empty?
-    cli.say '- Unregistered datasets detected: '
-    if notok.size <= 5
-      notok.keys.each { |i| cli.say "  > #{i}" }
-    else
-      cli.say "  > #{notok.size}, including #{notok.keys.first}"
-    end
-    cli.say '- Removing tables, recompute'
-    res.remove!
+    # TODO: Find different 95%ANI clusters with genomes from the same species
+    # TODO: Find AAI values too high or too low for each LCA rank
   end
 end