miga-base 0.3.1.7 → 0.3.2.0

data/lib/miga/common/path.rb

@@ -0,0 +1,49 @@
+
+module MiGA::Common::Path
+
+  ##
+  # Root path to MiGA (as estimated from the location of the current file).
+  def root_path
+    File.expand_path('../../../..', __FILE__)
+  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 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
+
+end
+
+##
+# MiGA extensions to the File class.
+class File
+
+  ##
+  # Method to transfer a file from +old_name+ to +new_name+, using a +method+
+  # that can be one of :symlink for File#symlink, :hardlink for File#link, or
+  # :copy for FileUtils#cp_r.
+  def self.generic_transfer(old_name, new_name, method)
+    return nil if exist? new_name
+    if(method==:copy)
+      FileUtils.cp_r(old_name, new_name)
+    else
+      method=:link if method==:hardlink
+      File.send(method, old_name, new_name)
+    end
+  end
+
+end
+

data/lib/miga/daemon.rb

@@ -2,13 +2,14 @@
 # @license Artistic-2.0
 
 require 'miga/project'
-require 'daemons'
-require 'date'
+require 'miga/daemon/base'
 
 ##
 # MiGA Daemons handling job submissions.
 class MiGA::Daemon < MiGA::MiGA
 
+  include MiGA::Daemon::Base
+
   ##
   # When was the last time a daemon for the MiGA::Project +project+ was seen
   # active? Returns DateTime.
@@ -61,64 +62,6 @@ class MiGA::Daemon < MiGA::MiGA
       multiple: false, log_output: true }
   end
 
-  ##
-  # Set/get #options, where +k+ is the Symbol of the option and +v+ is the value
-  # (or nil to use as getter). Skips consistency tests if +force+. Returns new
-  # value.
-  def runopts(k, v=nil, force=false)
-    k = k.to_sym
-    unless v.nil?
-      if [:latency, :maxjobs, :ppn].include?(k)
-        v = v.to_i
-      elsif [:shutdown_when_done].include?(k)
-        v = !!v
-      end
-      raise "Daemon's #{k} cannot be set to zero." if !force and v==0
-      @runopts[k] = v
-    end
-    if k==:kill and v.nil?
-      case @runopts[:type].to_s
-      when 'bash' then return "kill -9 '%s'"
-      when 'qsub' then return "qdel '%s'"
-      else             return "canceljob '%s'"
-      end
-    end
-    @runopts[k]
-  end
-
-  ##
-  # Returns Integer indicating the number of seconds to sleep between checks.
-  def latency() runopts(:latency); end
-
-  ##
-  # Returns Integer indicating the maximum number of concurrent jobs to run.
-  def maxjobs() runopts(:maxjobs); end
-
-  ##
-  # Returns Integer indicating the number of CPUs per job.
-  def ppn() runopts(:ppn); end
-
-  ##
-  # Returns Boolean indicating if the daemon should shutdown when processing is
-  # complete.
-  def shutdown_when_done?() !!runopts(:shutdown_when_done); end
-
-  ##
-  # Initializes the daemon with +opts+.
-  def start(opts=[]) daemon('start', opts); end
-
-  ##
-  # Stops the daemon with +opts+.
-  def stop(opts=[]) daemon('stop', opts); end
-
-  ##
-  # Restarts the daemon with +opts+.
-  def restart(opts=[]) daemon('restart', opts); end
-
-  ##
-  # Returns the status of the daemon with +opts+.
-  def status(opts=[]) daemon('status', opts); end
-
   ##
   # Launches the +task+ with options +opts+ (as command-line arguments).
   # Supported tasks include: start, stop, restart, status.

data/lib/miga/daemon/base.rb

@@ -0,0 +1,69 @@
+
+require 'daemons'
+require 'date'
+
+class MiGA::Daemon < MiGA::MiGA
+end
+
+module MiGA::Daemon::Base
+
+  ##
+  # Set/get #options, where +k+ is the Symbol of the option and +v+ is the value
+  # (or nil to use as getter). Skips consistency tests if +force+. Returns new
+  # value.
+  def runopts(k, v=nil, force=false)
+    k = k.to_sym
+    unless v.nil?
+      if [:latency, :maxjobs, :ppn].include?(k)
+        v = v.to_i
+      elsif [:shutdown_when_done].include?(k)
+        v = !!v
+      end
+      raise "Daemon's #{k} cannot be set to zero." if !force and v==0
+      @runopts[k] = v
+    end
+    if k==:kill and v.nil?
+      case @runopts[:type].to_s
+      when 'bash' then return "kill -9 '%s'"
+      when 'qsub' then return "qdel '%s'"
+      else             return "canceljob '%s'"
+      end
+    end
+    @runopts[k]
+  end
+
+  ##
+  # Returns Integer indicating the number of seconds to sleep between checks.
+  def latency() runopts(:latency); end
+
+  ##
+  # Returns Integer indicating the maximum number of concurrent jobs to run.
+  def maxjobs() runopts(:maxjobs); end
+
+  ##
+  # Returns Integer indicating the number of CPUs per job.
+  def ppn() runopts(:ppn); end
+
+  ##
+  # Returns Boolean indicating if the daemon should shutdown when processing is
+  # complete.
+  def shutdown_when_done?() !!runopts(:shutdown_when_done); end
+
+  ##
+  # Initializes the daemon with +opts+.
+  def start(opts=[]) daemon('start', opts); end
+
+  ##
+  # Stops the daemon with +opts+.
+  def stop(opts=[]) daemon('stop', opts); end
+
+  ##
+  # Restarts the daemon with +opts+.
+  def restart(opts=[]) daemon('restart', opts); end
+
+  ##
+  # Returns the status of the daemon with +opts+.
+  def status(opts=[]) daemon('status', opts); end
+
+end
+

data/lib/miga/dataset.rb

@@ -1,9 +1,9 @@
 # @package MiGA
 # @license Artistic-2.0
 
-require "miga/metadata"
-require "miga/dataset/result"
-require "sqlite3"
+require 'miga/metadata'
+require 'miga/dataset/result'
+require 'sqlite3'
 
 ##
 # Dataset representation in MiGA.

data/lib/miga/dataset/result.rb

@@ -1,7 +1,7 @@
 
-require "sqlite3"
-require "miga/result"
-require "miga/dataset/base"
+require 'sqlite3'
+require 'miga/result'
+require 'miga/dataset/base'
 
 ##
 # Helper module including specific functions to add dataset results.
@@ -35,8 +35,8 @@ module MiGA::Dataset::Result
   # returns a result if the expected files are complete. The +opts+ hash
   # controls result creation (if necessary). Supported values include:
   # - +is_clean+: A Boolean indicating if the input files are clean.
-  # - +force+: A Boolean indicating if the result must be re-indexed. If true, it
-  #   implies save=true.
+  # - +force+: A Boolean indicating if the result must be re-indexed. If true,
+  #   it implies save=true.
   # Returns MiGA::Result or nil.
   def add_result(result_type, save=true, opts={})
     dir = @@RESULT_DIRS[result_type]

data/lib/miga/result.rb

@@ -117,6 +117,11 @@ class MiGA::Result < MiGA::MiGA
   # Save the result persistently (in the JSON file #path).
   def save
     @data[:updated] = Time.now.to_s
+    s = path(:start)
+    if File.exist? s
+      @data[:started] = File.read(s).chomp
+      File.unlink s
+    end
     json = JSON.pretty_generate data
     ofh = File.open(path, "w")
     ofh.puts json

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 = [0.3, 1, 7]
+  VERSION = [0.3, 2, 0]
   
   ##
   # Nickname for the current major.minor version.
@@ -22,15 +22,17 @@ module MiGA
   
   ##
   # Reference of MiGA.
-  CITATION = "Rodriguez-R et al, in preparation. Microbial Genomes Atlas: " +
-    "Standardizing genomic and metagenomic analyses for Archaea and Bacteria."
+  CITATION = "Rodriguez-R et al (2018). " +
+    "The Microbial Genomes Atlas (MiGA) webserver: taxonomic and gene " +
+    "diversity analysis of Archaea and Bacteria at the whole genome level. " +
+    "Nucleic Acids Research, gky467. DOI: 10.1093/nar/gky467."
 
 end
 
 class MiGA::MiGA
-  
+
   include MiGA
-  
+
   ##
   # Major.minor version as Float.
   def self.VERSION ; VERSION[0] ; end

data/scripts/distances.bash

@@ -10,26 +10,9 @@ cd "$PROJECT/data/09.distances"
 
 # Initialize
 miga date > "$DATASET.start"
-TMPDIR=$(mktemp -d /tmp/MiGA.XXXXXXXXXXXX)
-trap "rm -rf '$TMPDIR'; exit" SIGHUP SIGINT SIGTERM
 
-# Check type of dataset
-NOMULTI=$(miga ls -P "$PROJECT" -D "$DATASET" --no-multi \
-  | wc -l | awk '{print $1}')
-REF=$(miga ls -P "$PROJECT" -D "$DATASET" --ref \
-  | wc -l | awk '{print $1}')
-
-# Call submodules
-# shellcheck source=scripts/_distances_functions.bash
-source "$MIGA/scripts/_distances_functions.bash"
-if [[ "$NOMULTI" -eq "1" && "$REF" -eq "1" ]] ; then
-  # shellcheck source=scripts/_distances_ref_nomulti.bash
-  source "$MIGA/scripts/_distances_ref_nomulti.bash"
-elif [[ "$NOMULTI" -eq "1" ]] ; then
-  S_PROJ=$PROJECT
-  # shellcheck source=scripts/_distances_noref_nomulti.bash
-  source "$MIGA/scripts/_distances_noref_nomulti.bash"
-fi
+# Run
+ruby -I "$MIGA/lib" "$MIGA/utils/distances.rb" "$PROJECT" "$DATASET"
 
 # Finalize
 rm -R "$TMPDIR"

data/scripts/taxonomy.bash

@@ -13,27 +13,8 @@ cd "$DIR"
 # Initialize
 miga date > "$DATASET.start"
 
-# Check if there is a reference project
-S_PROJ=$(miga about -P "$PROJECT" -m ref_project)
-
-if [[ "$S_PROJ" != "?" ]] ; then
-
-  # Check type of dataset
-  NOMULTI=$(miga ls -P "$PROJECT" -D "$DATASET" --no-multi \
-    | wc -l | awk '{print $1}')
-
-  if [[ "$NOMULTI" -eq "1" ]] ; then
-    # Call submodules
-    TMPDIR=$(mktemp -d /tmp/MiGA.XXXXXXXXXXXX)
-    trap "rm -rf '$TMPDIR'; exit" SIGHUP SIGINT SIGTERM
-    # shellcheck source=scripts/_distances_functions.bash
-    source "$MIGA/scripts/_distances_functions.bash"
-    # shellcheck source=scripts/_distances_noref_nomulti.bash
-    source "$MIGA/scripts/_distances_noref_nomulti.bash"
-    rm -R "$TMPDIR"
-  fi
-
-fi
+# Run
+ruby "$MIGA/utils/distances.rb" "$PROJECT" "$DATASET" run_taxonomy=1
 
 # Finalize
 miga date > "$DATASET.done"

data/test/common_test.rb

@@ -79,4 +79,13 @@ class CommonTest < Test::Unit::TestCase
     assert_equal("678  90", tab[3])
   end
 
+  def test_miga_name
+    assert_equal('Xa sp. C', 'Xa_sp__C'.unmiga_name)
+    assert_equal('X_______', 'X^*.!{}!'.miga_name)
+    assert_equal('aB09', 'aB09'.miga_name)
+    assert('R2D2'.miga_name?)
+    assert(!'C3-PO'.miga_name?)
+    assert_equal("123\n1\n", '1231'.wrap_width(3))
+  end
+
 end

data/utils/distance/base.rb

@@ -0,0 +1,6 @@
+
+require 'miga'
+require 'miga/tax_dist'
+
+class MiGA::DistanceRunner
+end

data/utils/distance/commands.rb

@@ -0,0 +1,82 @@
+
+module MiGA::DistanceRunner::Commands
+  # Estimates or calculates AAI against +target+
+  def aai(target)
+    # Check if the request makes sense
+    return nil if target.nil? or target.result(:essential_genes).nil?
+    # Check if it's been calculated
+    y = stored_value(target, :aai)
+    return y unless y.nil? or y.zero?
+    # Try hAAI (except in clade projects)
+    unless @ref_project.is_clade?
+      y = haai(target)
+      return y unless y.nil? or y.zero?
+    end
+    # Full AAI
+    aai_cmd(
+          tmp_file("proteins.fa"), target.result(:cds).file_path(:proteins),
+          dataset.name, target.name, tmp_dbs[:aai]).tap{ checkpoint :aai }
+  end
+
+  # Estimates AAI against +target+ using hAAI
+  def haai(target)
+    haai = aai_cmd(tmp_file("ess_genes.fa"),
+          target.result(:essential_genes).file_path(:ess_genes),
+          dataset.name, target.name, tmp_dbs[:haai],
+          aai_save_rbm: "no-save-rbm", aai_p: opts[:haai_p])
+    checkpoint :haai
+    return nil if haai.nil? or haai.zero? or haai > 90.0
+    aai = 100.0 - Math.exp(2.435076 + 0.4275193*Math.log(100.0-haai))
+    SQLite3::Database.new(tmp_dbs[:aai]) do |conn|
+      conn.execute "insert into aai values(?, ?, ?, 0, 0, 0)",
+            [dataset.name, target.name, aai]
+    end
+    aai
+  end
+
+  # Calculates ANI against +target+
+  def ani(target)
+    # Check if the request makes sense
+    t = tmp_file("largecontigs.fa")
+    r = target.result(:assembly)
+    return nil if r.nil? or !File.size?(t)
+    # Check if it's been calculated
+    y = stored_value(target, :ani)
+    return y unless y.nil? or y.zero?
+    # Run it
+    ani_cmd(
+        t, r.file_path(:largecontigs),
+        dataset.name, target.name, tmp_dbs[:ani]).tap{ checkpoint :ani }
+  end
+
+  # Execute an AAI command
+  def aai_cmd(f1, f2, n1, n2, db, o={})
+    o = opts.merge(o)
+    v = `aai.rb -1 "#{f1}" -2 "#{f2}" -S "#{db}" \
+          --name1 "#{n1}" --name2 "#{n2}" \
+          -t "#{o[:thr]}" -a --lookup-first "--#{o[:aai_save_rbm]}" \
+          -p "#{o[:aai_p] || "blast+"}"`.chomp
+    (v.nil? or v.empty?) ? 0 : v.to_f
+  end
+
+  # Execute an ANI command
+  def ani_cmd(f1, f2, n1, n2, db, o={})
+    o = opts.merge(o)
+    v = nil
+    if o[:ani_p] == 'fastani'
+      out = `fastANI -r "#{f1}" -q "#{f2}" \
+            -o /dev/stdout 2>/dev/null`.chomp.split(/\s+/)
+      SQLite3::Database.new(db) do |conn|
+        conn.execute "insert into ani values(?, ?, ?, 0, ?, ?)",
+              [n1, n2, out[2], out[3], out[4]]
+      end
+      v = out[2]
+    else
+      v = `ani.rb -1 "#{f1}" -2 "#{f2}" -S "#{db}" \
+            --name1 "#{n1}" --name2 "#{n2}" \
+            -t "#{opts[:thr]}" -a --no-save-regions --no-save-rbm \
+            --lookup-first -p "#{o[:ani_p] || "blast+"}"`.chomp
+    end
+    v.nil? or v.empty? ? 0 : v.to_f
+  end
+end