miga-base 0.7.22.0 → 0.7.25.0

data/lib/miga/project/result.rb

@@ -31,9 +31,9 @@ module MiGA::Project::Result
   ##
   # Is this +task+ to be bypassed?
   def ignore_task?(task)
-    metadata["run_#{task}"] == false ||
-      (!is_clade? && @@INCLADE_TASKS.include?(task) &&
-        metadata["run_#{task}"] != true)
+    return true if metadata["run_#{task}"] == false
+
+    !clade? && @@INCLADE_TASKS.include?(task) && metadata["run_#{task}"] != true
   end
 
   ##
@@ -74,7 +74,7 @@ module MiGA::Project::Result
       return r
     end
     return nil unless result_files_exist?(base, %w[.proposed-clades])
-    unless is_clade? ||
+    unless clade? ||
            result_files_exist?(
              base, %w[.pdf .classif .medoids .class.tsv .class.nwk]
            )

data/lib/miga/remote_dataset.rb

@@ -162,6 +162,8 @@ class MiGA::RemoteDataset < MiGA::MiGA
       txt.empty? ? sleep(1) : break
     end
     doc = MiGA::Json.parse(txt, symbolize: false, contents: true)
+    return if doc.nil? || doc['result'].nil? || doc['result'].empty?
+
     @_ncbi_asm_json_doc = doc['result'][ doc['result']['uids'].first ]
   end
 

data/lib/miga/result/stats.rb

@@ -118,7 +118,7 @@ module MiGA::Result::Stats
 
   def compute_stats_essential_genes
     stats = {}
-    if source.is_multi?
+    if source.multi?
       stats = { median_copies: 0, mean_copies: 0 }
       File.open(file_path(:report), 'r') do |fh|
         fh.each_line do |ln|
@@ -151,7 +151,7 @@ module MiGA::Result::Stats
       source.save
 
       # Inactivate low-quality datasets
-      min_qual = (project.metadata[:min_qual] || 25)
+      min_qual = project.option(:min_qual)
       if min_qual != 'no' && stats[:quality] < min_qual
         source.inactivate! 'Low quality genome'
       end

data/lib/miga/version.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'date'
 
 ##
@@ -8,7 +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 = [0.7, 22, 0]
+  VERSION = [0.7, 25, 0].freeze
 
   ##
   # Nickname for the current major.minor version.
@@ -16,7 +18,7 @@ module MiGA
 
   ##
   # Date of the current gem release.
-  VERSION_DATE = Date.new(2021, 1, 23)
+  VERSION_DATE = Date.new(2021, 2, 22)
 
   ##
   # Reference of MiGA.

data/scripts/essential_genes.bash

@@ -24,18 +24,33 @@ fi
 mkdir "${DATASET}.ess"
 TYPE=$(miga ls -P "$PROJECT" -D "$DATASET" \
   --metadata "type" | awk '{print $2}')
-COLL=$(miga about -P "$PROJECT" -m ess_coll)
-[[ "$COLL" == "?" ]] && COLL=dupont_2012
+COLL=$(miga option -P "$PROJECT" --key ess_coll)
 if [[ "$TYPE" == "metagenome" || "$TYPE" == "virome" ]] ; then
   FLAGS="--metagenome"
 else
-  FLAGS="--alignments ${DATASET}.ess/proteins.aln"
+  FLAGS=""
 fi
 HMM.essential.rb \
   -i "$FAA" -o "${DATASET}.ess.faa" -m "${DATASET}.ess/" \
   -t "$CORES" -r "$DATASET" --collection "$COLL" $FLAGS \
   > "${DATASET}.ess/log"
 
+# Index for FastAAI
+NOMULTI=$(miga list_datasets -P "$PROJECT" -D "$DATASET" --no-multi \
+            | wc -l | awk '{print $1}')
+if [[ "$NOMULTI" -eq "1" ]] ; then
+  if [[ "$FAA" == *.gz ]] ; then
+    gzip -cd "$FAA" > "${DATASET}.faix"
+  else
+    cp "$FAA" "${DATASET}.faix"
+  fi
+  FastAAI --qp "${DATASET}.faix" --output "${DATASET}.faix" \
+    --ext ".faix" --index --input-paths --all-vs-all --threads "$CORES"
+  rm "${DATASET}.faix"
+  rm "${DATASET}.faix.hmm"
+  rm "${DATASET}.faix.hmm.filt"
+fi
+
 # Reduce files
 if exists "$DATASET".ess/*.faa ; then
   ( cd "${DATASET}.ess" \

data/scripts/miga.bash

@@ -1,12 +1,17 @@
 #!/bin/bash
+
+# Setup environment
 set -e
-#MIGA=${MIGA:-$(cd "$(dirname "$0")/.."; pwd)}
 MIGA_HOME=${MIGA_HOME:-"$HOME"}
+SCRIPT=${SCRIPT:-$(basename "$0" .bash)}
 # shellcheck source=/dev/null
 . "$MIGA_HOME/.miga_rc"
+
+# Ensure submodules are first in PATH
 export PATH="$MIGA/bin:$MIGA/utils/enveomics/Scripts:$PATH"
-SCRIPT=${SCRIPT:-$(basename "$0" .bash)}
+export PATH="$MIGA/utils/FastAAI/FastAAI:$PATH"
 
+# Ancillary functions
 function exists { [[ -e "$1" ]] ; }
 function fx_exists { [[ $(type -t "$1") == "function" ]] ; }
 function miga_start_project_step {
@@ -28,6 +33,7 @@ function miga_end_project_step {
   miga add_result -P "$PROJECT" -r "$SCRIPT" -f
 }
 
+# Environment header
 if [[ "$SCRIPT" != "d" && "$SCRIPT" != "p" ]] ; then
   echo ""
   echo "######[ $SCRIPT ]######"

data/scripts/mytaxa.bash

@@ -14,15 +14,16 @@ if [[ "$MIGA_MYTAXA" == "no" ]] ; then
   echo "This system doesn't currently support MyTaxa." \
     > "$DATASET.nomytaxa.txt"
 else
-  MT=$(dirname -- "$(which MyTaxa)")
-
   # Check type of dataset
   MULTI=$(miga list_datasets -P "$PROJECT" -D "$DATASET" --multi \
     | wc -l | awk '{print $1}')
   if [[ "$MULTI" -eq "1" ]] ; then
     # Check requirements
-    if [[ ! -e "$MT/AllGenomes.faa.dmnd" ]] ; then
-      echo "Cannot locate the database: $MT/AllGenomes.faa.dmnd:" \
+    MT=$(dirname -- "$(which MyTaxa)")
+    DB="$MIGA_HOME/.miga_db/AllGenomes.faa.dmnd"
+    [[ -e "$DB" ]] || DB="$MT/AllGenomes.faa.dmnd"
+    if [[ ! -e "$DB" ]] ; then
+      echo "Cannot locate the database: AllGenomes.faa.dmnd:" \
             "no such file or directory" >&2
       exit 1
     fi
@@ -40,7 +41,7 @@ else
     # Execute search
     FAA="../../../06.cds/$DATASET.faa"
     [[ -s "$FAA" ]] || FAA="${FAA}.gz"
-    diamond blastp -q "$FAA" -d "$MT/AllGenomes.faa" \
+    diamond blastp -q "$FAA" -d "$DB" \
       -a "$DATASET.daa" -k 5 -p "$CORES" --min-score 60
     diamond view -a "$DATASET.daa" -o "$DATASET.blast"
 

data/scripts/mytaxa_scan.bash

@@ -5,7 +5,6 @@ SCRIPT="mytaxa_scan"
 # shellcheck source=scripts/miga.bash
 . "$MIGA/scripts/miga.bash" || exit 1
 DIR="$PROJECT/data/07.annotation/03.qa/02.mytaxa_scan"
-[[ -d "$DIR" ]] || mkdir -p "$DIR"
 cd "$DIR"
 
 # Initialize
@@ -14,17 +13,16 @@ if [[ "$MIGA_MYTAXA" == "no" ]] ; then
   echo "This system doesn't currently support MyTaxa." \
     > "$DATASET.nomytaxa.txt"
 else
-  MT=$(dirname -- "$(which MyTaxa)")
-  TMPDIR=$(mktemp -d /tmp/MiGA.XXXXXXXXXXXX)
-  trap "rm -rf '$TMPDIR'; exit" SIGHUP SIGINT SIGTERM
-
   # Check type of dataset
   NOMULTI=$(miga list_datasets -P "$PROJECT" -D "$DATASET" --no-multi \
     | wc -l | awk '{print $1}')
   if [[ "$NOMULTI" -eq "1" ]] ; then
     # Check requirements
+    MT=$(dirname -- "$(which MyTaxa)")
+    DB="$MIGA_HOME/.miga_db/AllGenomes.faa.dmnd"
+    [[ -e "$DB" ]] || DB="$MT/AllGenomes.faa.dmnd"
     if [[ ! -e "$MT/AllGenomes.faa.dmnd" ]] ; then
-      echo "Cannot locate the database: $MT/AllGenomes.faa.dmnd:" \
+      echo "Cannot locate the database: AllGenomes.faa.dmnd:" \
             "no such file or directory" >&2
       exit 1
     fi
@@ -39,13 +37,16 @@ else
       exit 1
     fi
      
+    TMPDIR=$(mktemp -d /tmp/MiGA.XXXXXXXXXXXX)
+    trap "rm -rf '$TMPDIR'; exit" SIGHUP SIGINT SIGTERM
+
     FAA="../../../06.cds/$DATASET.faa"
     [[ -s "$FAA" ]] || FAA="${FAA}.gz"
     if [[ ! -s "$DATASET.mytaxa" ]] ; then
       # Execute search
       if [[ ! -s "$DATASET.blast" ]] ; then
         diamond blastp -q "$FAA" -a "$DATASET.daa" -t "$TMPDIR" \
-          -d "$MT/AllGenomes.faa" -k 5 -p "$CORES" --min-score 60
+          -d "$DB" -k 5 -p "$CORES" --min-score 60
         diamond view -a "$DATASET.daa" -o "$DATASET.blast" -t "$TMPDIR"
       fi
 

data/scripts/ogs.bash

@@ -12,8 +12,7 @@ miga_start_project_step "$DIR"
 DS=$(miga ls -P "$PROJECT" --ref --no-multi)
 
 if [[ -n $DS ]] ; then
-  MIN_ID=$(miga about -P "$PROJECT" -m ogs_identity)
-  [[ $MIN_ID == "?" ]] && MIN_ID=80
+  MIN_ID=$(miga option -P "$PROJECT" --key ogs_identity)
   if [[ ! -s miga-project.ogs ]] ; then
     # Extract RBMs
     if [[ ! -s miga-project.abc ]] ; then
@@ -34,7 +33,7 @@ if [[ -n $DS ]] ; then
 
     # Estimate OGs and Clean RBMs
     ogs.mcl.rb -o miga-project.ogs --abc miga-project.abc -t "$CORES"
-    if [[ $(miga about -P "$PROJECT" -m clean_ogs) == "false" ]] ; then
+    if [[ $(miga option -P "$PROJECT" --key clean_ogs) == "false" ]] ; then
       gzip -9 miga-project.abc
     else
       rm miga-project.abc

data/scripts/ssu.bash

@@ -16,6 +16,7 @@ if [[ -s $fa ]] ; then
   # Run barrnap
   barrnap --quiet --threads "$CORES" "$fa" | grep "^##gff\\|;product=16S " \
     > "$DATASET.ssu.gff"
+
   # Extract
   bedtools getfasta -s "-fi" "$fa" -bed "$DATASET.ssu.gff" \
     -fo "$DATASET.ssu.all.fa"
@@ -24,9 +25,22 @@ if [[ -s $fa ]] ; then
   FastA.filter.pl "$DATASET.ssu.fa.id" "$DATASET.ssu.all.fa" > "$DATASET.ssu.fa"
   rm "$DATASET.ssu.fa.id"
   [[ -e "$fa.fai" ]] && rm "$fa.fai"
+
+  # RDP classifier
+  if [[ "$MIGA_RDP" == "yes" && -s "$DATASET.ssu.all.fa" ]] ; then
+    java -jar "$MIGA_HOME/.miga_db/classifier.jar" classify \
+      -c 0.8 -f fixrank -g 16srrna -o "$DATASET.rdp.tsv" \
+      "$DATASET.ssu.all.fa"
+    echo "# Version: $(perl -pe 's/.*://' \
+          < "$MIGA_HOME/.miga_db/classifier.version.txt" \
+          | grep . | paste - - | perl -pe 's/\t/; /')" \
+      >> "$DATASET.rdp.tsv"
+  fi
+
   # Gzip
-  gzip -9 -f "$DATASET.ssu.gff"
-  gzip -9 -f "$DATASET.ssu.all.fa"
+  for x in ssu.gff ssu.all.fa rdp.tsv ; do
+    [[ -e "${DATASET}.${x}" ]] && gzip -9 -f "${DATASET}.${x}"
+  done
 fi
 
 # Finalize

data/test/dataset_test.rb

@@ -29,7 +29,7 @@ class DatasetTest < Test::Unit::TestCase
     assert_raise { MiGA::Dataset.new(project, 'dataset-1') }
     assert_equal(project, dataset.project)
     assert_equal('dataset0', dataset.name)
-    assert_predicate(dataset, :is_ref?)
+    assert_predicate(dataset, :ref?)
     assert_equal(MiGA::Metadata, dataset.metadata.class)
     assert_equal(:incomplete, dataset.status)
   end
@@ -38,14 +38,14 @@ class DatasetTest < Test::Unit::TestCase
     d2 = project.add_dataset('ds_save')
     assert_respond_to(d2, :save)
     d2.save
-    assert_not_predicate(d2, :is_multi?)
-    assert_not_predicate(d2, :is_nonmulti?)
+    assert_not_predicate(d2, :multi?)
+    assert_not_predicate(d2, :nonmulti?)
     assert_nil(d2.metadata[:type])
     d2.metadata[:type] = :metagenome
     d2.save
     assert_equal(:metagenome, d2.metadata[:type])
-    assert_predicate(d2, :is_multi?)
-    assert_not_predicate(d2, :is_nonmulti?)
+    assert_predicate(d2, :multi?)
+    assert_not_predicate(d2, :nonmulti?)
   end
 
   def test_remove

data/test/lair_test.rb

@@ -75,10 +75,9 @@ class LairTest < Test::Unit::TestCase
       assert_predicate(d, :active?)
     end
 
-    out = capture_stderr { lair.terminate_daemons }.string
+    out = capture_stderr { lair.terminate_daemons ; sleep(2) }.string
     assert_match(/Probing MiGA::Daemon/, out)
     assert_match(/Sending termination message/, out)
-    sleep(2)
     assert_not_predicate(d, :active?)
 
     out = capture_stderr { assert { lair.daemon_loop } }.string

data/test/net_test.rb

@@ -0,0 +1,34 @@
+require 'test_helper'
+
+class FormatTest < Test::Unit::TestCase
+  include TestHelper
+
+  def test_known_hosts
+    m = MiGA::MiGA
+    assert_not_nil(m.known_hosts(:miga_db))
+    assert_not_nil(m.known_hosts('miga_db'))
+    assert_not_nil(m.known_hosts(:miga_dist))
+    assert_raise { m.known_kosts(:not_a_host) }
+  end
+
+  def test_remote_connection
+    declare_remote_access
+    m = MiGA::MiGA
+    assert_raise { m.remote_connection(:bad_descriptor) }
+    assert_raise { m.remote_connection('http://microbial-genomes.org/') }
+    c = m.remote_connection(:miga_db)
+    assert_equal(Net::FTP, c.class)
+    c.close
+  end
+
+  def test_download_file_ftp
+    declare_remote_access
+    m = MiGA::MiGA
+    f = tmpfile('t/test.txt')
+    d = File.dirname(f)
+    assert(!Dir.exist?(d))
+    m.download_file_ftp(:miga_online_ftp, 'test.txt', f)
+    assert(Dir.exist?(d))
+    assert_equal('miga', File.read(f).chomp)
+  end
+end

data/test/with_option_test.rb

@@ -0,0 +1,115 @@
+# frozen_string_literal: true
+
+require 'test_helper'
+require 'miga/common/with_option'
+
+class WithDaemonTest < Test::Unit::TestCase
+  include TestHelper
+
+  class TestWithOption < MiGA::MiGA
+    include MiGA::Common::WithOption
+
+    attr_reader :metadata, :saved
+
+    def initialize
+      @metadata = { range: 0.9 }
+      @saved = false
+    end
+
+    def self.OPTIONS
+      {
+        empty: {},
+        float: { type: Float },
+        range: { default: 1.0, in: -5.5..5.5, type: Float },
+        default: { default: 9, type: Integer },
+        token: { type: Integer, tokens: %w[yes no 0] },
+        proc: { default: proc { Date.today } },
+        bool: { in: [true, false] }
+      }
+    end
+
+    def save
+      @saved = true
+    end
+  end
+
+  def test_with_option
+    o = TestWithOption.new
+    assert_respond_to(o, :option)
+    assert_equal(1, o.metadata.size)
+  end
+
+  def test_option
+    o = TestWithOption.new
+    assert_equal(9, o.option(:default))
+    assert_nil(o.option(:bool))
+    assert_raise { o.option(:not_an_option) }
+    assert_nil(o.option(:empty))
+  end
+
+  def test_set_bool
+    o = TestWithOption.new
+    assert_nil(o.option(:bool))
+    assert(!o.saved)
+    assert_raise { o.set_option(:bool, 'true') }
+    assert_nil(o.option(:bool))
+    assert(!o.saved)
+    assert_equal(true, o.set_option(:bool, 'true', true))
+    assert(o.saved)
+    assert_equal(false, o.set_option(:bool, false))
+    assert_equal(false, o.set_option(:bool, 'false', true))
+    assert_nil(o.set_option(:bool, nil))
+  end
+
+  def test_set_empty
+    o = TestWithOption.new
+    assert_nil(o.option(:empty))
+    assert_equal('a', o.set_option(:empty, 'a'))
+    assert_equal('1', o.set_option(:empty, '1', true))
+  end
+
+  def test_all_options
+    o = TestWithOption.new
+    assert(o.all_options.is_a?(Hash))
+    assert_include(o.all_options.keys, :bool)
+    assert_nil(o.all_options[:bool])
+  end
+
+  def test_option?
+    o = TestWithOption.new
+    assert(o.option?(:range))
+    assert(!o.option?(:not_an_option))
+  end
+
+  def test_option_metadata
+    o = TestWithOption.new
+    assert_equal(0.9, o.option(:range))
+    assert_equal(1.0, o.set_option(:range, nil))
+    assert_equal(2.0, o.set_option(:range, 2.0))
+    assert_equal(3.0, o.set_option(:range, '3', true))
+  end
+
+  def test_option_range
+    o = TestWithOption.new
+    assert_raise { o.set_option(:range, 9.0) }
+    assert_raise { o.set_option(:range, 3) }
+    assert_raise { o.set_option(:range, true) }
+  end
+
+  def test_option_proc
+    o = TestWithOption.new
+    assert(o.option(:proc).is_a?(Date))
+    assert(o.set_option(:proc, 1).is_a?(Integer))
+    assert(o.set_option(:proc, nil).is_a?(Date))
+  end
+
+  def test_token
+    o = TestWithOption.new
+    assert_nil(o.option(:token))
+    assert_equal(1, o.set_option(:token, 1))
+    assert_equal(-2, o.set_option(:token, '-2', true))
+    assert_equal('yes', o.set_option(:token, 'yes'))
+    assert_equal('0', o.set_option(:token, '0', true))
+    assert_raise { o.set_option(:token, 'maybe') }
+  end
+end