miga-base 0.2.6.4 → 0.2.6.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 378d848c9b2abcbaad34e80b4ddf17308f2ff3e8
-  data.tar.gz: 7a56bf5ef16a43b6e3ec44b3658efb85151aca9e
+  metadata.gz: a69a12511e98d2cc6751efa31592661b664e9d33
+  data.tar.gz: 807b0a13efce0367c88369a2f389fffff34ee58d
 SHA512:
-  metadata.gz: fd8270bb64964cd340f237c92d922650d39fedbb098993f73225f5de9ee29948d8bf27024157fc64850a1d6fb3b192313c2bb5861fae6aeca8b693acab705b3a
-  data.tar.gz: 788a60e1807a65593f81030c8bab14ffa5f9924ae69101381a1c7d77b3f5329c94eb365496dcbd04c71155066e010c4c1338b074776c5b7023c101a1ec424806
+  metadata.gz: bdc1db4adec179da57c45ab29a41ec36ef2895f34d8159792420560bd0bc15910052fe6e2949ee0a992500e0001a7db4d2bbbf4f4d5e1daefb4c1c0fa3bd4ebb
+  data.tar.gz: 40d5a05ec2c154b50eb8952b87212106461080740e8986f982f44fa79cea3100d31beaff1c647afa83892b3e2cde37464ddd11568ce658addcb60992b6395de3

data/actions/list_datasets.rb

@@ -3,7 +3,7 @@
 # @package MiGA
 # @license Artistic-2.0
 
-o = {q:true, info:false, processing:false}
+o = {q:true, info:false, processing:false, silent:false}
 OptionParser.new do |opt|
   opt_banner(opt)
   opt_object(opt, o, [:project, :dataset_opt])
@@ -15,11 +15,15 @@ OptionParser.new do |opt|
   opt.on("-m", "--metadata STRING",
     "Print name and metadata field only. If set, ignores -i."
     ){ |v| o[:datum]=v }
+  opt.on("-s", "--silent",
+    "No output and exit with non-zero status if the dataset list is empty."
+    ){ |v| o[:silent] = v }
   opt_common(opt, o)
 end.parse!
 
 ##=> Main <=
 opt_require(o, project:"-P")
+o[:q] = true if o[:silent]
 
 $stderr.puts "Loading project." unless o[:q]
 p = MiGA::Project.load(o[:project])
@@ -34,6 +38,7 @@ else
   ds = []
 end
 ds = filter_datasets!(ds, o)
+exit(1) if o[:silent] and ds.empty?
 
 if not o[:datum].nil?
   ds.each{|d| puts "#{d.name}\t#{d.metadata[ o[:datum] ] || "?"}"}

data/actions/run_local.rb

@@ -36,7 +36,7 @@ else
 end
 raise "Unsupported #{type.to_s.gsub(/.*::/,"")} result: #{o[:name]}." if
   type.RESULT_DIRS[o[:name].to_sym].nil?
-cmd << "#{miga}/scripts/#{o[:name]}.bash".shellescape
+cmd << MiGA::MiGA.script_path(o[:name], miga:miga, project:p).shellescape
 pid = spawn cmd.join(" ")
 Process.wait pid
 

data/actions/tax_distributions.rb

@@ -38,13 +38,13 @@ raise "#{res_n} not yet calculated." if res.nil?
 matrix = res.file_path(:matrix)
 raise "#{res_n} has no matrix." if matrix.nil?
 dist = {}
-fh = matrix=~/\.gz$/ ? Zlib::GzipReader.open(matrix) : File.open(matrix,"r")
-fh.each_line do |ln|
-  next if fh.lineno==1
+mfh = matrix=~/\.gz$/ ? Zlib::GzipReader.open(matrix) : File.open(matrix,"r")
+mfh.each_line do |ln|
+  next if mfh.lineno==1
   row = ln.chomp.split(/\t/)
   dist[cannid(row[1], row[2])] = [row[3], 0, ["root:biota"]]
 end
-fh.close
+mfh.close
 
 Dir.mktmpdir do |dir|
   if o[:index].nil?

data/lib/miga/common.rb

@@ -103,6 +103,24 @@ class MiGA::MiGA
       tmp.unlink
     end
   end
+  
+  ##
+  # Path to a script to be executed for +task+. Supported +opts+ are:
+  # - +:miga+ Path to the MiGA home to use. If not passed, the home of the
+  #   library is used).
+  # - +:project+ MiGA::Project object to check within plugins. If not passed,
+  #   only core scripts are supported.
+  def self.script_path(task, opts={})
+    opts[:miga] ||= root_path
+    unless opts[:project].nil?
+      opts[:project].plugins.each do |pl|
+        if File.exist? File.expand_path("scripts/#{task}.bash", pl)
+          opts[:miga] = pl
+        end
+      end
+    end
+    File.expand_path("scripts/#{task}.bash", opts[:miga])
+  end
 
 
   ##

data/lib/miga/daemon.rb

@@ -167,7 +167,7 @@ class MiGA::Daemon < MiGA::MiGA
     to_run = {ds: ds, job: job, task_name: task_name,
       cmd: sprintf(runopts(:cmd),
         # 1: script
-        File.expand_path("scripts/#{job}.bash", vars["MIGA"]),
+        MiGA::MiGA.script_path(job, miga:vars["MIGA"], project:project),
         # 2: vars
         vars.keys.map { |k|
 	  sprintf(runopts(:var), k, vars[k]) }.join(runopts(:varsep)),

data/lib/miga/dataset_result.rb

@@ -27,24 +27,25 @@ module MiGA::DatasetResult
   private
 
     ##
-    # Add result type +:raw_reads+ at +base+ (no +opts+ supported).
-    def add_result_raw_reads(base, opts)
+    # Add result type +:raw_reads+ at +base+ (no +_opts+ supported).
+    def add_result_raw_reads(base, _opts)
       return nil unless result_files_exist?(base, ".1.fastq")
       r = MiGA::Result.new("#{base}.json")
-      r = add_files_to_ds_result(r, name,
+      add_files_to_ds_result(r, name,
         ( result_files_exist?(base, ".2.fastq") ?
-          {:pair1=>".1.fastq", :pair2=>".2.fastq"} :
-          {:single=>".1.fastq"} ))
+          {pair1:".1.fastq", pair2:".2.fastq"} :
+          {single:".1.fastq"} ))
     end
 
     ##
-    # Add result type +:trimmed_reads+ at +base+ (no +opts+ supported).
-    def add_result_trimmed_reads(base, opts)
+    # Add result type +:trimmed_reads+ at +base+ (no +_opts+ supported).
+    def add_result_trimmed_reads(base, _opts)
       return nil unless result_files_exist?(base, ".1.clipped.fastq")
       r = MiGA::Result.new("#{base}.json")
-      r = add_files_to_ds_result(r, name,
-        {:pair1=>".1.clipped.fastq", :pair2=>".2.clipped.fastq"}) if
-        result_files_exist?(base, ".2.clipped.fastq")
+      if result_files_exist?(base, ".2.clipped.fastq")
+        r = add_files_to_ds_result(r, name,
+          pair1:".1.clipped.fastq", pair2:".2.clipped.fastq")
+      end
       r.add_file(:single, "#{name}.1.clipped.single.fastq")
       r.add_file(:trimming_sumary, "#{name}.1.fastq.trimmed.summary.txt")
       add_result(:raw_reads) #-> Post gunzip
@@ -52,26 +53,26 @@ module MiGA::DatasetResult
     end
 
     ##
-    # Add result type +:read_quality+ at +base+ (no +opts+ supported).
-    def add_result_read_quality(base, opts)
+    # Add result type +:read_quality+ at +base+ (no +_opts+ supported).
+    def add_result_read_quality(base, _opts)
       return nil unless result_files_exist?(base, %w[.solexaqa .fastqc])
       r = MiGA::Result.new("#{base}.json")
       r = add_files_to_ds_result(r, name,
-        {:solexaqa=>".solexaqa", :fastqc=>".fastqc"})
+        solexaqa:".solexaqa", fastqc:".fastqc")
       add_result(:trimmed_reads) #-> Post cleaning
       r
     end
 
     ##
-    # Add result type +:trimmed_fasta+ at +base+ (no +opts+ supported).
-    def add_result_trimmed_fasta(base, opts)
+    # Add result type +:trimmed_fasta+ at +base+ (no +_opts+ supported).
+    def add_result_trimmed_fasta(base, _opts)
       return nil unless
         result_files_exist?(base, ".CoupledReads.fa") or
         result_files_exist?(base, ".SingleReads.fa") or
         result_files_exist?(base, %w[.1.fasta .2.fasta])
       r = MiGA::Result.new("#{base}.json")
-      r = add_files_to_ds_result(r, name, {:coupled=>".CoupledReads.fa",
-        :single=>".SingleReads.fa", :pair1=>".1.fasta", :pair2=>".2.fasta"})
+      r = add_files_to_ds_result(r, name, coupled:".CoupledReads.fa",
+        single:".SingleReads.fa", pair1:".1.fasta", pair2:".2.fasta")
       add_result(:raw_reads) #-> Post gzip
       r
     end
@@ -82,8 +83,8 @@ module MiGA::DatasetResult
     def add_result_assembly(base, opts)
       return nil unless result_files_exist?(base, ".LargeContigs.fna")
       r = MiGA::Result.new("#{base}.json")
-      r = add_files_to_ds_result(r, name, {:largecontigs=>".LargeContigs.fna",
-        :allcontigs=>".AllContigs.fna", :assembly_data=>""})
+      r = add_files_to_ds_result(r, name, largecontigs:".LargeContigs.fna",
+        allcontigs:".AllContigs.fna", assembly_data:"")
       opts[:is_clean] ||= false
       r.clean! if opts[:is_clean]
       unless r.clean?
@@ -99,8 +100,8 @@ module MiGA::DatasetResult
     def add_result_cds(base, opts)
       return nil unless result_files_exist?(base, %w[.faa .fna])
       r = MiGA::Result.new("#{base}.json")
-      r = add_files_to_ds_result(r, name, {:proteins=>".faa", :genes=>".fna",
-        :gff2=>".gff2", :gff3=>".gff3", :tab=>".tab"})
+      r = add_files_to_ds_result(r, name, proteins:".faa", genes:".fna",
+        gff2:".gff2", gff3:".gff3", tab:".tab")
       opts[:is_clean] ||= false
       r.clean! if opts[:is_clean]
       unless r.clean?
@@ -112,12 +113,12 @@ module MiGA::DatasetResult
     end
 
     ##
-    # Add result type +:essential_genes+ at +base+ (no +opts+ supported).
-    def add_result_essential_genes(base, opts)
+    # Add result type +:essential_genes+ at +base+ (no +_opts+ supported).
+    def add_result_essential_genes(base, _opts)
       return nil unless result_files_exist?(base, %w[.ess.faa .ess .ess/log])
       r = MiGA::Result.new("#{base}.json")
-      r = add_files_to_ds_result(r, name, {:ess_genes=>".ess.faa",
-        :collection=>".ess", :report=>".ess/log"})
+      add_files_to_ds_result(r, name, ess_genes:".ess.faa",
+        collection:".ess", report:".ess/log")
     end
 
     ##
@@ -126,8 +127,8 @@ module MiGA::DatasetResult
       return MiGA::Result.new("#{base}.json") if result(:assembly).nil?
       return nil unless result_files_exist?(base, ".ssu.fa")
       r = MiGA::Result.new("#{base}.json")
-      r = add_files_to_ds_result(r, name, {:longest_ssu_gene=>".ssu.fa",
-        :gff=>".ssu.gff", :all_ssu_genes=>".ssu.all.fa"})
+      r = add_files_to_ds_result(r, name, longest_ssu_gene:".ssu.fa",
+        gff:".ssu.gff", all_ssu_genes:".ssu.all.fa")
       opts[:is_clean] ||= false
       r.clean! if opts[:is_clean]
       unless r.clean?
@@ -138,37 +139,36 @@ module MiGA::DatasetResult
     end
 
     ##
-    # Add result type +:mytaxa+ at +base+ (no +opts+ supported).
-    def add_result_mytaxa(base, opts)
+    # Add result type +:mytaxa+ at +base+ (no +_opts+ supported).
+    def add_result_mytaxa(base, _opts)
       if is_multi?
         return nil unless result_files_exist?(base, ".mytaxa")
         r = MiGA::Result.new("#{base}.json")
-        add_files_to_ds_result(r, name, {:mytaxa=>".mytaxa", :blast=>".blast",
-          :mytaxain=>".mytaxain"})
+        add_files_to_ds_result(r, name, mytaxa:".mytaxa", blast:".blast",
+          mytaxain:".mytaxain")
       else
         MiGA::Result.new("#{base}.json")
       end
     end
 
     ##
-    # Add result type +:mytaxa_scan+ at +base+ (no +opts+ supported).
-    def add_result_mytaxa_scan(base, opts)
+    # Add result type +:mytaxa_scan+ at +base+ (no +_opts+ supported).
+    def add_result_mytaxa_scan(base, _opts)
       if is_nonmulti?
         return nil unless
           result_files_exist?(base, %w[.pdf .wintax .mytaxa .reg])
         r = MiGA::Result.new("#{base}.json")
-        add_files_to_ds_result(r, name, {:mytaxa=>".mytaxa", :wintax=>".wintax",
-          :blast=>".blast", :mytaxain=>".mytaxain", :report=>".pdf",
-          :regions=>".reg", :gene_ids=>".wintax.genes",
-          :region_ids=>".wintax.regions"})
+        add_files_to_ds_result(r, name, mytaxa:".mytaxa", wintax:".wintax",
+          blast:".blast", mytaxain:".mytaxain", report:".pdf", regions:".reg",
+          gene_ids:".wintax.genes", region_ids:".wintax.regions")
       else
         MiGA::Result.new("#{base}.json")
       end
     end
 
     ##
-    # Add result type +:distances+ at +base+ (no +opts+ supported).
-    def add_result_distances(base, opts)
+    # Add result type +:distances+ at +base+ (no +_opts+ supported).
+    def add_result_distances(base, _opts)
       if is_nonmulti?
         if is_ref?
           add_result_distances_ref(base)
@@ -181,8 +181,8 @@ module MiGA::DatasetResult
     end
 
     ##
-    # Add result type +:stats+ at +base+ (no +opts+ supported).
-    def add_result_stats(base, opts)
+    # Add result type +:stats+ at +base+ (no +_opts+ supported).
+    def add_result_stats(base, _opts)
       MiGA::Result.new("#{base}.json")
     end
     
@@ -199,8 +199,8 @@ module MiGA::DatasetResult
       return nil unless
         File.exist?("#{pref}/01.haai/#{name}.db")
       r = MiGA::Result.new("#{base}.json")
-      r.add_files({:haai_db=>"01.haai/#{name}.db",
-        :aai_db=>"02.aai/#{name}.db", :ani_db=>"03.ani/#{name}.db"})
+      r.add_files(haai_db:"01.haai/#{name}.db", aai_db:"02.aai/#{name}.db",
+        ani_db:"03.ani/#{name}.db")
       r
     end
 
@@ -211,10 +211,9 @@ module MiGA::DatasetResult
         result_files_exist?(base, %w[.aai-medoids.tsv .aai.db]) or
         result_files_exist?(base, %w[.ani-medoids.tsv .ani.db])
       r = MiGA::Result.new("#{base}.json")
-      r = add_files_to_ds_result(r, name, {
-        :aai_medoids=>".aai-medoids.tsv",
-        :haai_db=>".haai.db", :aai_db=>".aai.db",
-        :ani_medoids=>".ani-medoids.tsv", :ani_db=>".ani.db"})
+      add_files_to_ds_result(r, name, aai_medoids:".aai-medoids.tsv",
+        haai_db:".haai.db", aai_db:".aai.db", ani_medoids:".ani-medoids.tsv",
+        ani_db:".ani.db", ref_tree:".nwk", ref_tree_pdf:".nwk.pdf")
     end
 
     ##

data/lib/miga/remote_dataset.rb

@@ -58,26 +58,9 @@ class MiGA::RemoteDataset < MiGA::MiGA
     ids = [ids] unless ids.is_a? Array
     case @@UNIVERSE[universe][:method]
     when :rest
-      map_to = @@UNIVERSE[universe][:dbs][db].nil? ? nil :
-        @@UNIVERSE[universe][:dbs][db][:map_to]
-      url = sprintf @@UNIVERSE[universe][:url],
-        db, ids.join(","), format, map_to
-      response = RestClient::Request.execute(method: :get, url:url, timeout:600)
-      raise "Unable to reach #{universe} client, error code " +
-        "#{response.code}." unless response.code == 200
-      doc = response.to_s
+      doc = download_rest(universe, db, ids, format)
     when :net
-      url = sprintf @@UNIVERSE[universe][:url],db,ids.join(","),format,map_to
-      doc = ""
-      @timeout_try = 0
-      begin
-        open(url) { |f| doc = f.read }
-      rescue Net::ReadTimeout
-        @timeout_try += 1
-        if @timeout_try > 3 ; raise Net::ReadTimeout
-        else ; retry
-        end
-      end
+      doc = download_net(universe, db, ids, format)
     end
     unless file.nil?
       ofh = File.open(file, "w")
@@ -87,6 +70,38 @@ class MiGA::RemoteDataset < MiGA::MiGA
     doc
   end
 
+  ##
+  # Download data usint a REST method from the +universe+ in the database +db+
+  # with IDs +ids+ and in +format+. Returns the doc as String.
+  def self.download_rest(universe, db, ids, format)
+    u = @@UNIVERSE[universe]
+    map_to = u[:dbs][db].nil? ? nil : u[:dbs][db][:map_to]
+    url = sprintf(u[:url], db, ids.join(","), format, map_to)
+    response = RestClient::Request.execute(method: :get, url:url, timeout:600)
+    unless response.code == 200
+      raise "Unable to reach #{universe} client, error code #{response.code}."
+    end
+    response.to_s
+  end
+  
+  ##
+  # Download data usint a REST method from the +universe+ in the database +db+
+  # with IDs +ids+ and in +format+. Returns the doc as String.
+  def self.download_net(universe, db, ids, format)
+    url = sprintf(@@UNIVERSE[universe][:url], db, ids.join(","), format, map_to)
+    doc = ""
+    @timeout_try = 0
+    begin
+      open(url) { |f| doc = f.read }
+    rescue Net::ReadTimeout
+      @timeout_try += 1
+      if @timeout_try > 3 ; raise Net::ReadTimeout
+      else ; retry
+      end
+    end
+    doc
+  end
+
   # Instance-level
 
   ##
@@ -119,33 +134,37 @@ class MiGA::RemoteDataset < MiGA::MiGA
   # Save dataset to the MiGA::Project +project+ identified with +name+. +is_ref+
   # indicates if it should be a reference dataset, and contains +metadata+.
   def save_to(project, name=nil, is_ref=true, metadata={})
-    name = ids.join("_").miga_name if name.nil?
+    name ||= ids.join("_").miga_name
     project = MiGA::Project.new(project) if project.is_a? String
-    raise "Dataset #{name} exists in the project, aborting..." if
-      MiGA::Dataset.exist?(project, name)
+    if MiGA::Dataset.exist?(project, name)
+      raise "Dataset #{name} exists in the project, aborting..."
+    end
     metadata = get_metadata(metadata)
-    case @@UNIVERSE[universe][:dbs][db][:stage]
+    udb = @@UNIVERSE[universe][:dbs][db]
+    metadata["#{universe}_#{db}"] = ids.join(",")
+    case udb[:stage]
     when :assembly
       dir = MiGA::Dataset.RESULT_DIRS[:assembly]
       base = "#{project.path}/data/#{dir}/#{name}"
+      l_ctg = "#{base}.LargeContigs.fna"
+      a_ctg = "#{base}.AllContigs.fna"
       File.open("#{base}.start", "w") { |ofh| ofh.puts Time.now.to_s }
-      if @@UNIVERSE[universe][:dbs][db][:format] == :fasta_gz
-        download("#{base}.LargeContigs.fna.gz")
-        system("gzip -d #{base}.LargeContigs.fna.gz")
+      if udb[:format] == :fasta_gz
+        download "#{l_ctg}.gz"
+        system "gzip -d '#{l_ctg}.gz'"
       else
-        download("#{base}.LargeContigs.fna")
+        download l_ctg
       end
-      File.symlink(
-        File.basename("#{base}.LargeContigs.fna"), "#{base}.AllContigs.fna")
+      File.unlink(a_ctg) if File.exist? a_ctg
+      File.symlink(File.basename(l_ctg), a_ctg)
       File.open("#{base}.done", "w") { |ofh| ofh.puts Time.now.to_s }
     else
       raise "Unexpected error: Unsupported result for database #{db}."
     end
     dataset = MiGA::Dataset.new(project, name, is_ref, metadata)
     project.add_dataset(dataset.name)
-    result = dataset.add_result(@@UNIVERSE[universe][:dbs][db][:stage],
-      true, is_clean:true)
-    raise "Empty dataset created: seed result was not added due to "+
+    result = dataset.add_result(udb[:stage], true, is_clean:true)
+    raise "Empty dataset created: seed result was not added due to " +
       "incomplete files." if result.nil?
     result.clean!
     result.save
@@ -212,4 +231,5 @@ class MiGA::RemoteDataset < MiGA::MiGA
       return nil unless ln =~ /^\d+$/
       ln
     end
+
 end

data/lib/miga/tax_dist.rb

@@ -25,7 +25,7 @@ module MiGA::TaxDist
     Zlib::GzipReader.open(aai_path(test)) do |fh|
       keys = nil
       fh.each_line do |ln|
-        row = ln.chomp.split /\t/
+        row = ln.chomp.split(/\t/)
         if fh.lineno==1
           keys = row[1, row.size-1].map{ |i| i.to_i }
         elsif row.shift.to_f >= aai
@@ -56,7 +56,7 @@ module MiGA::TaxDist
       min = pv.values.select{ |v| v < upr }.max
       return out if min.nil?
       if min >= lwr
-        v = pv.select{ |_,v| v==min }
+        v = pv.select{ |_,vj| vj==min }
         out[phrase] = (test==:intax ? v.reverse_each : v).first
       end
     end

data/lib/miga/tax_index.rb

@@ -57,7 +57,7 @@ class MiGA::TaxIndex < MiGA::MiGA
   ##
   # Generate JSON String for the index.
   def to_json
-    JSON.pretty_generate({ root:root.to_hash,
+    JSON.generate({ root:root.to_hash,
       datasets:datasets.map{ |d| d.name } })
   end
 

data/lib/miga/version.rb

@@ -10,7 +10,7 @@ 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 = [0.2, 6, 4]
+  VERSION = [0.2, 6, 5]
   
   ##
   # Nickname for the current major.minor version.
@@ -18,7 +18,7 @@ module MiGA
   
   ##
   # Date of the current gem release.
-  VERSION_DATE = Date.new(2017, 4, 15)
+  VERSION_DATE = Date.new(2017, 5, 22)
   
   ##
   # Reference of MiGA.

data/scripts/_distances_functions.bash

@@ -28,8 +28,10 @@ fx_exists miga-aai || function miga-aai {
   local F2=$2
   local TH=$3
   local DB=$4
-  local N1=$(miga-ds_name "$F1")
-  local N2=$(miga-ds_name "$F2")
+  local N1
+  N1=$(miga-ds_name "$F1")
+  local N2
+  N2=$(miga-ds_name "$F2")
   aai.rb -1 "$F1" -2 "$F2" -t "$TH" -a --lookup-first -S "$DB" --name1 "$N1" \
     --name2 "$N2" --$MIGA_AAI_SAVE_RBM || echo "0"
 }
@@ -39,8 +41,10 @@ fx_exists miga-ani || function miga-ani {
   local F2=$2
   local TH=$3
   local DB=$4
-  local N1=$(miga-ds_name "$F1")
-  local N2=$(miga-ds_name "$F2")
+  local N1
+  N1=$(miga-ds_name "$F1")
+  local N2
+  N2=$(miga-ds_name "$F2")
   ani.rb -1 "$F1" -2 "$F2" -t "$TH" -a --no-save-regions --no-save-rbm \
     --lookup-first -S "$DB" --name1 "$N1" --name2 "$N2" || echo "0"
 }
@@ -51,11 +55,15 @@ fx_exists miga-haai || function miga-haai {
   local TH=$3
   local DB=$4
   local AAI_DB=$5
-  local N1=$(miga-ds_name "$F1")
-  local N2=$(miga-ds_name "$F2")
-  local HAAI=$(MIGA_AAI_SAVE_RBM="no-save-rbm" miga-aai "$F1" "$F2" "$TH" "$DB")
+  local N1
+  N1=$(miga-ds_name "$F1")
+  local N2
+  N2=$(miga-ds_name "$F2")
+  local HAAI
+  HAAI=$(MIGA_AAI_SAVE_RBM="no-save-rbm" miga-aai "$F1" "$F2" "$TH" "$DB")
   if [[ "$HAAI" != "" && $(perl -e "print 1 if '$HAAI' <= 90") == "1" ]] ; then
-    local AAI=$(perl -e "print (100-exp(2.435076 + 0.4275193*log(100-$HAAI)))")
+    local AAI
+    AAI=$(perl -e "print (100-exp(2.435076 + 0.4275193*log(100-$HAAI)))")
     [[ ! -s $AAI_DB ]] && miga-make_empty_aai_db "$AAI_DB"
     echo "insert into aai values('$N1','$N2','$AAI',0,0,0);" | sqlite3 "$AAI_DB"
     echo "$AAI"
@@ -70,6 +78,7 @@ fx_exists miga-haai_or_aai || function miga-haai_or_aai {
   local  F2=$5
   local  DB=$6
   local  TH=$7
+  local AAI
   AAI=$(miga-haai "$FH1" "$FH2" "$TH" "$DBH" "$DB")
   [[ "${AAI%.*}" -le 0 ]] && AAI=$(miga-aai "$F1" "$F2" "$TH" "$DB")
   echo "$AAI"

data/scripts/_distances_noref_nomulti.bash

@@ -36,24 +36,24 @@ fx_exists miga-noref_haai_or_aai || function miga-noref_haai_or_aai {
 fx_exists miga-noref_ani || function miga-noref_ani {
   local Q=$1
   local S=$2
-  [[ -s $TMPDIR/$Q.LargeContigs.fna ]] \
-    || cp ../05.assembly/$Q.LargeContigs.fna "$TMPDIR/$Q.LargeContigs.fna"
+  [[ -s "$TMPDIR/$Q.LargeContigs.fna" ]] \
+    || cp "../05.assembly/$Q.LargeContigs.fna" "$TMPDIR/$Q.LargeContigs.fna"
   miga-ani "$TMPDIR/$Q.LargeContigs.fna" "../05.assembly/$S.LargeContigs.fna" \
     "$CORES" "$TMPDIR/$Q.ani.db"
 }
 
-
-
 # Calculate the classification-informed AAI/ANI traverse (if not classified)
 ESS="../07.annotation/01.function/01.essential"
 if [[ $(miga project_info -P "$PROJECT" -m type) != "clade" ]] ; then
   # Classify aai-clade (if project type is not clade)
   CLADES="../10.clades/01.find"
   METRIC="aai"
+  REF_TABLE="02.aai/miga-project.txt.gz"
 else
   # Classify ani-clade (if project type is clade)
   CLADES="../10.clades/02.ani"
   METRIC="ani"
+  REF_TABLE="03.ani/miga-project.txt.gz"
 fi
 
 CLASSIF="."
@@ -63,7 +63,7 @@ while [[ -e "$CLADES/$CLASSIF/miga-project.medoids" ]] ; do
   VAL_MED=""
   VAL_CLS=""
   i_n=0
-  while read i ; do
+  while read -r i ; do
     let i_n=$i_n+1
     if [[ $METRIC == "aai" ]] ; then
       VAL=$(miga-noref_haai_or_aai "$DATASET" "$i")
@@ -87,7 +87,7 @@ done
 if [[ "$CLASSIF" != "." ]] ; then
   PAR=$(dirname "$CLADES/$CLASSIF")/miga-project.classif
   if [[ -s "$PAR" ]] ; then
-    while read i ; do
+    while read -r i ; do
       if [[ $METRIC == "aai" ]] ; then
         AAI=$(miga-noref_haai_or_aai "$DATASET" "$i")
       else
@@ -101,6 +101,25 @@ if [[ "$CLASSIF" != "." ]] ; then
   fi
 fi
 
-#Finalize
+# Finalize
 N=11
 miga-checkpoint_n
+
+# Build tree with medoids
+if [[ -s "${DATASET}.${METRIC}.db" ]] ; then
+  echo "select seq2 from $METRIC;" | sqlite3 "${DATASET}.${METRIC}.db" \
+    | sort | uniq > "${DATASET}.tmp0"
+  perl -pe "s/^/^/" "${DATASET}.tmp0" | perl -pe "s/$/\\t/" \
+    > "${DATASET}.tmp1"
+  perl -pe "s/^/\\t/" "${DATASET}.tmp0" | perl -pe "s/$/\\t/" \
+    > "${DATASET}.tmp2"
+  echo "a b value" | tr " " "\\t" > "${DATASET}.txt"
+  gzip -c -d "$REF_TABLE" | cut -f 2-4 \
+    | grep -f "${DATASET}.tmp1" | grep -f "${DATASET}.tmp2" \
+    >> "${DATASET}.txt"
+  echo "select seq1, seq2, $METRIC from $METRIC;" \
+    | sqlite3 "${DATASET}.${METRIC}.db"  | tr "\\|" "\\t" \
+    >> "${DATASET}.txt"
+  "$MIGA/utils/ref-tree.R" "${DATASET}.txt" "$DATASET" "$DATASET"
+  rm "$DATASET".tmp[012] "${DATASET}.txt"
+fi

data/scripts/aai_distances.bash

@@ -4,11 +4,12 @@ set -e
 SCRIPT="aai_distances"
 echo "MiGA: $MIGA"
 echo "Project: $PROJECT"
+# shellcheck source=scripts/miga.bash
 source "$MIGA/scripts/miga.bash" || exit 1
 cd "$PROJECT/data/09.distances/02.aai"
 
 # Initialize
-date "+%Y-%m-%d %H:%M:%S %z" > "miga-project.start"
+miga date > "miga-project.start"
 
 echo -n "" > miga-project.log
 DS=$(miga list_datasets -P "$PROJECT" --ref --no-multi)
@@ -40,5 +41,5 @@ if(sum(aai[,'a'] != aai[,'b']) > 0){
 gzip -9 -f miga-project.txt
 
 # Finalize
-date "+%Y-%m-%d %H:%M:%S %z" > "miga-project.done"
+miga date > "miga-project.done"
 miga add_result -P "$PROJECT" -r "$SCRIPT"

data/scripts/ani_distances.bash

@@ -4,11 +4,12 @@ set -e
 SCRIPT="ani_distances"
 echo "MiGA: $MIGA"
 echo "Project: $PROJECT"
+# shellcheck source=scripts/miga.bash
 source "$MIGA/scripts/miga.bash" || exit 1
 cd "$PROJECT/data/09.distances/03.ani"
 
 # Initialize
-date "+%Y-%m-%d %H:%M:%S %z" > "miga-project.start"
+miga date > "miga-project.start"
 
 echo -n "" > miga-project.log
 DS=$(miga list_datasets -P "$PROJECT" --ref --no-multi)
@@ -39,5 +40,5 @@ if(sum(ani[,'a'] != ani[,'b']) > 0){
 gzip -9 -f miga-project.txt
 
 # Finalize
-date "+%Y-%m-%d %H:%M:%S %z" > "miga-project.done"
+miga date > "miga-project.done"
 miga add_result -P "$PROJECT" -r "$SCRIPT"