miga-base 0.2.2.1 → 0.2.2.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: c38a1e0856fa157209b6f14f61a77f2fa2f8effa
-  data.tar.gz: b10f449bb2ae56b2956207188a5147b10912b847
+  metadata.gz: 569b4e7cff6063f23d5c76445ce5fe8379c152bd
+  data.tar.gz: 9ea1a8bff35874d39a27eb0544e4f9b262441368
 SHA512:
-  metadata.gz: ca6f0ed0b982074fa85ccb43f9bdfcda36a4efd6431a9ea34ddcc6cc0a7c67cb66863071648175246c80aadee774ab5c6206ae3cda85178ce4f9bb55468ab72a
-  data.tar.gz: 52b06e6702717d44f603c7d4c0e4cef5e00e5704b6d9ed283808d79fa765b10db384aa60ceda9383114a9e23f56a8ebf000b6c9818f1cc29e261c7759c4861e5
+  metadata.gz: 0eda211778880b2dcaca0b5504ce560aeed3f7f4d420e0c127019ce86f68511e67af329aa920ab0816460c8730eed8cb0f46f6b12debf637ce8e95891d670cf0
+  data.tar.gz: 22d7016022c1d63d848b8046ca8899dbf95bf22fa132613783f9c1b72ca0e289e1ce653a5b300824eeb64697354bbdbc705950db8b147604e38318e0456811d6

data/Gemfile

@@ -1,3 +1,6 @@
 source "https://rubygems.org"
 gemspec name: "miga-base"
-gem "codeclimate-test-reporter", group: :test, require: nil
+group :test do
+  gem "simplecov"
+  gem "codeclimate-test-reporter", "~> 1.0.0"
+end

data/actions/create_dataset.rb

@@ -56,11 +56,8 @@ end
   d.metadata[k]=o[k] unless o[k].nil?
 end
 
-if o[:update]
-  d.save
-else
-  p.add_dataset(o[:dataset])
-end
+d.save
+p.add_dataset(o[:dataset]) unless o[:update]
 res = d.first_preprocessing(true)
 $stderr.puts "- #{res}" unless o[:q]
 

data/actions/daemon.rb

@@ -36,6 +36,7 @@ OptionParser.new do |opt|
   opt.on("-f", "--force", "Force operation"){ o[:daemon_opts] << '-f' }
   opt.on("-n", "--no_wait",
     "Do not wait for processes to stop"){ o[:daemon_opts] << '-n' }
+  opt.on("--shush", "Silence the daemon."){ o[:daemon_opts] << '--shush' }
 end.parse!
 
 ##=> Main <=

data/actions/plugins.rb

@@ -0,0 +1,25 @@
+#!/usr/bin/env ruby
+
+# @package MiGA
+# @license Artistic-2.0
+
+o = {q:true, update:false}
+OptionParser.new do |opt|
+  opt_banner(opt)
+  opt_object(opt, o, [:project])
+  opt.on("--install PATH",
+    "Installs the specified plugin in the project."){ |v| o[:install]=v }
+  opt.on("--uninstall PATH",
+    "Uninstalls the specified plugin from the project."){ |v| o[:uninstall]=v }
+  opt_common(opt, o)
+end.parse!
+
+##=> Main <=
+opt_require(o, project:"-P")
+
+p = MiGA::Project.new(o[:project], true)
+p.install_plugin(o[:install]) unless o[:install].nil?
+p.uninstall_plugin(o[:uninstall]) unless o[:uninstall].nil?
+p.plugins.each { |i| puts i }
+
+$stderr.puts "Done." unless o[:q]

data/actions/result_stats.rb

@@ -59,6 +59,16 @@ if o[:compute]
     f = r.file_path :proteins
     s = `FastA.length.pl '#{f}' | #{scr}`.chomp.split(" ")
     stats = {predicted_proteins: s[0].to_i, average_length: [s[1].to_f, "aa"]}
+  when :essential_genes
+    stats = {completeness:[0.0,"%"], contamination:[0.0,"%"]}
+    File.open(r.file_path(:report), "r") do |fh|
+      fh.each_line do |ln|
+        if /^! (Completeness|Contamination): (.*)%/.match(ln)
+          stats[$1.downcase.to_sym][0] = $2.to_f
+        end
+      end
+    end
+    stats[:quality] = stats[:completeness][0] - stats[:contamination][0]*5
   else
     stats = nil
   end

data/bin/miga

@@ -14,6 +14,7 @@ $task_desc = {
   # Projects
   create_project: "Creates an empty MiGA project.",
   project_info: "Displays information about a MiGA project.",
+  plugins: "Lists or (un)installs plugins in a MiGA project.",
   # Datasets
   create_dataset: "Creates an empty dataset in a pre-existing MiGA project.",
   download_dataset: "Creates an empty dataset in a pre-existing MiGA project.",

data/lib/miga/daemon.rb

@@ -202,14 +202,22 @@ class MiGA::Daemon < MiGA::MiGA
     while jobs_running.size < maxjobs
       break if jobs_to_run.empty?
       job = @jobs_to_run.shift
+      # Launch job
       if runopts(:type) == "bash"
         job[:pid] = spawn job[:cmd]
-        Process.detach job[:pid]
+        Process.detach job[:pid] unless job[:pid].nil? or job[:pid].empty?
       else
         job[:pid] = `#{job[:cmd]}`.chomp
       end
-      @jobs_running << job
-      say "Spawned pid:#{job[:pid]} for #{job[:task_name]}."
+      # Check if registered
+      if job[:pid].nil? or job[:pid].empty?
+        job[:pid] = nil
+        @jobs_to_run << job
+        say "Unsuccessful #{job[:task_name]}, rescheduling."
+      else
+        @jobs_running << job
+        say "Spawned pid:#{job[:pid]} for #{job[:task_name]}."
+      end
     end
   end
 
@@ -232,6 +240,7 @@ class MiGA::Daemon < MiGA::MiGA
     end
     @loop_i += 1
     declare_alive
+    project.load
     check_datasets
     check_project
     flush!
@@ -239,7 +248,6 @@ class MiGA::Daemon < MiGA::MiGA
       say "Housekeeping for sanity"
       @loop_i = 0
       purge!
-      project.load
     end
     sleep(latency)
   end

data/lib/miga/dataset.rb

@@ -106,8 +106,8 @@ class MiGA::Dataset < MiGA::MiGA
     @project = project
     @name = name
     metadata[:ref] = is_ref
-    @metadata = MiGA::Metadata.new(project.path + "/metadata/" + name + ".json",
-      metadata)
+    @metadata = MiGA::Metadata.new(
+      File.expand_path("metadata/#{name}.json", project.path), metadata )
   end
   
   ##
@@ -181,7 +181,8 @@ class MiGA::Dataset < MiGA::MiGA
     return nil if @@RESULT_DIRS[result_type].nil?
     base = project.path + "/data/" + @@RESULT_DIRS[result_type] +
       "/" + name
-    return MiGA::Result.load("#{base}.json") unless save
+    r_pre = MiGA::Result.load("#{base}.json")
+    return r_pre unless r_pre.nil? or save
     return nil unless result_files_exist?(base, ".done")
     r = self.send("add_result_#{result_type}", base)
     r.save unless r.nil?

data/lib/miga/project.rb

@@ -77,8 +77,7 @@ class MiGA::Project < MiGA::MiGA
   ##
   # Project-wide tasks for :clade projects.
   def self.INCLADE_TASKS ; @@INCLADE_TASKS ; end
-  @@INCLADE_TASKS = [:subclades, :ogs, :ess_phylogeny, :core_phylogeny,
-    :clade_metadata]
+  @@INCLADE_TASKS = [:subclades, :ogs]
   
   ##
   # Does the project at +path+ exist?
@@ -110,8 +109,9 @@ class MiGA::Project < MiGA::MiGA
   def initialize(path, update=false)
     @datasets = {}
     @path = File.absolute_path(path)
-    self.create if update or not Project.exist? self.path
+    self.create if not update and not Project.exist? self.path
     self.load if self.metadata.nil?
+    self.load_plugins
   end
 
   ##
@@ -141,6 +141,7 @@ class MiGA::Project < MiGA::MiGA
   ##
   # (Re-)load project data and metadata.
   def load
+    @datasets = {}
     @metadata = MiGA::Metadata.load "#{path}/miga.project.json"
     raise "Couldn't find project metadata at #{path}" if metadata.nil?
   end
@@ -290,7 +291,9 @@ class MiGA::Project < MiGA::MiGA
   def unregistered_datasets
     datasets = []
     MiGA::Dataset.RESULT_DIRS.values.each do |dir|
-      Dir.entries("#{path}/data/#{dir}").each do |file|
+      dir_p = "#{path}/data/#{dir}"
+      next unless Dir.exist? dir_p
+      Dir.entries(dir_p).each do |file|
         next unless
           file =~ %r{
             \.(fa(a|sta|stqc?)?|fna|solexaqa|gff[23]?|done|ess)(\.gz)?$
@@ -331,6 +334,37 @@ class MiGA::Project < MiGA::MiGA
     each_dataset { |ds| blk.call(ds.profile_advance) }
   end
 
+  ##
+  # Installs the plugin in the specified path.
+  def install_plugin(path)
+    abs_path = File.absolute_path(path)
+    raise "Plugin already installed in project: #{abs_path}." unless
+      metadata[:plugins].nil? or not metadata[:plugins].include?(abs_path)
+    raise "Malformed MiGA plugin: #{abs_path}." unless
+      File.exist?(File.expand_path("miga-plugin.json", abs_path))
+    self.metadata[:plugins] ||= []
+    self.metadata[:plugins] << abs_path
+    save
+  end
+
+  ##
+  # Uninstall the plugin in the specified path.
+  def uninstall_plugin(path)
+    abs_path = File.absolute_path(path)
+    raise "Plugin not currently installed: #{abs_path}." if
+      metadata[:plugins].nil? or not metadata[:plugins].include?(abs_path)
+    self.metadata[:plugins].delete(abs_path)
+    save
+  end
+
+  ##
+  # List plugins installed in the project.
+  def plugins ; metadata[:plugins] ||= [] ; end
+
+  def load_plugins
+    plugins.each { |pl| require File.expand_path("lib-plugin.rb", pl) }
+  end
+
   private
 
     ##

data/lib/miga/remote_dataset.rb

@@ -37,13 +37,13 @@ class MiGA::RemoteDataset < MiGA::MiGA
     },
     ncbi:{
       dbs: { nuccore:{stage: :assembly, format: :fasta} },
-      url: "http://eutils.ncbi.nlm.nih.gov/entrez/eutils/" +
+      url: "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/" +
         "efetch.fcgi?db=%1$s&id=%2$s&rettype=%3$s&retmode=text",
       method: :rest
     },
     ncbi_map:{
       dbs: { assembly:{map_to: :nuccore, format: :text} },
-      url: "http://eutils.ncbi.nlm.nih.gov/entrez/eutils/" +
+      url: "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/" +
         # FIXME ncbi_map is intended to do internal NCBI mapping between
         # databases.
         "elink.fcgi?dbfrom=%1$s&id=%2$s&db=%3$s - - - - -",

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, 2, 1]
+  VERSION = [0.2, 2, 2]
   
   ##
   # Nickname for the current major.minor version.

data/scripts/_distances_functions.bash

@@ -13,47 +13,47 @@ if [[ ! -n $MIGA_AAI_SAVE_RBM ]] ; then
   fi
 fi
 
-function make_empty_aai_db {
+fx_exists miga-make_empty_aai_db || function miga-make_empty_aai_db {
   local DB=$1
   echo "create table if not exists aai(seq1 varchar(256), seq2 varchar(256)," \
     " aai float, sd float, n int, omega int);" | sqlite3 $DB
 }
 
-function ds_name {
+fx_exists miga-ds_name || function miga-ds_name {
   basename $1 | perl -pe "s/[^A-Za-z0-9_].*//"
 }
 
-function aai {
+fx_exists miga-aai || function miga-aai {
   local F1=$1
   local F2=$2
   local TH=$3
   local DB=$4
-  local N1=$(ds_name $F1)
-  local N2=$(ds_name $F2)
+  local N1=$(miga-ds_name $F1)
+  local N2=$(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"
 }
 
-function ani {
+fx_exists miga-ani || function miga-ani {
   local F1=$1
   local F2=$2
   local TH=$3
   local DB=$4
-  local N1=$(ds_name $F1)
-  local N2=$(ds_name $F2)
+  local N1=$(miga-ds_name $F1)
+  local N2=$(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"
 }
 
-function haai {
+fx_exists miga-haai || function miga-haai {
   local F1=$1
   local F2=$2
   local TH=$3
   local DB=$4
   local AAI_DB=$5
-  local N1=$(ds_name $F1)
-  local N2=$(ds_name $F2)
-  local HAAI=$(MIGA_AAI_SAVE_RBM="no-save-rbm" aai $F1 $F2 $TH $DB)
+  local N1=$(miga-ds_name $F1)
+  local N2=$(miga-ds_name $F2)
+  local HAAI=$(MIGA_AAI_SAVE_RBM="no-save-rbm" miga-aai $F1 $F2 $TH $DB)
   if [[ "$HAAI" != "" && $(perl -e "print 1 if '$HAAI' <= 90") == "1" ]] ; then
     local AAI=$(perl -e "print (100-exp(2.435076 + 0.4275193*log(100-$HAAI)))")
     [[ ! -s $AAI_DB ]] && make_empty_aai_db $AAI_DB
@@ -62,7 +62,7 @@ function haai {
   fi
 }
 
-function haai_or_aai {
+fx_exists miga-haai_or_aai || function miga-haai_or_aai {
   local FH1=$1
   local FH2=$2
   local DBH=$3
@@ -70,12 +70,12 @@ function haai_or_aai {
   local  F2=$5
   local  DB=$6
   local  TH=$7
-  AAI=$(haai $FH1 $FH2 $TH $DBH $DB)
-  [[ "${AAI%.*}" -le 0 ]] && AAI=$(aai $F1 $F2 $TH $DB)
+  AAI=$(miga-haai $FH1 $FH2 $TH $DBH $DB)
+  [[ "${AAI%.*}" -le 0 ]] && AAI=$(miga-aai $F1 $F2 $TH $DB)
   echo $AAI
 }
 
-function val_from_db {
+fx_exists miga-val_from_db || function miga-val_from_db {
   local N1=$1
   local N2=$2
   local DB=$3
@@ -86,10 +86,10 @@ function val_from_db {
   fi
 }
 
-function aai_from_db {
-  val_from_db $1 $2 $3 aai
+fx_exists miga-aai_from_db || function miga-aai_from_db {
+  miga-val_from_db $1 $2 $3 aai
 }
 
-function ani_from_db {
-  val_from_db $1 $2 $3 ani
+fx_exists miga-ani_from_db || function miga-ani_from_db {
+  miga-val_from_db $1 $2 $3 ani
 }

data/scripts/_distances_noref_nomulti.bash

@@ -9,7 +9,7 @@ exists $DATASET.haai.db && cp $DATASET.haai.db $TMPDIR
 exists $DATASET.a[an]i.db && cp $DATASET.a[an]i.db $TMPDIR
 exists $DATASET.a[an]i.9[05] && rm $DATASET.a[an]i.9[05]
 N=0
-function checkpoint_n {
+fx_exists miga-checkpoint_n || function miga-checkpoint_n {
   let N=$N+1
   if [[ $N -ge 10 ]] ; then
     for metric in haai aai ani ; do
@@ -24,20 +24,27 @@ function checkpoint_n {
   fi
 }
 
-function noref_haai_or_aai {
+fx_exists miga-noref_haai_or_aai || function miga-noref_haai_or_aai {
   local Q=$1
   local S=$2
-  haai_or_aai $ESS/$Q.ess.faa $ESS/$S.ess.faa $TMPDIR/$Q.haai.db \
-    ../06.cds/$Q.faa ../06.cds/$S.faa $TMPDIR/$Q.aai.db $CORES
+  [[ -s $TMPDIR/$Q.faa ]] \
+    || cp ../06.cds/$Q.faa $TMPDIR/$Q.faa
+  miga-haai_or_aai $ESS/$Q.ess.faa $ESS/$S.ess.faa $TMPDIR/$Q.haai.db \
+    $TMPDIR/$Q.faa ../06.cds/$S.faa $TMPDIR/$Q.aai.db $CORES
 }
 
-function noref_ani {
+fx_exists miga-noref_ani || function miga-noref_ani {
   local Q=$1
   local S=$2
-  ani ../05.assembly/$Q.LargeContigs.fna ../05.assembly/$S.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)
@@ -59,11 +66,11 @@ while [[ -e "$CLADES/$CLASSIF/miga-project.medoids" ]] ; do
   for i in $(cat "$CLADES/$CLASSIF/miga-project.medoids") ; do
     let i_n=$i_n+1
     if [[ $METRIC == "aai" ]] ; then
-      VAL=$(noref_haai_or_aai $DATASET $i)
+      VAL=$(miga-noref_haai_or_aai $DATASET $i)
     else
-      VAL=$(noref_ani $DATASET $i)
+      VAL=$(miga-noref_ani $DATASET $i)
     fi
-    checkpoint_n
+    miga-checkpoint_n
     if [[ $(perl -e "print 1 if '$VAL' >= '$MAX_VAL'") == "1" ]] ; then
       MAX_VAL=$VAL
       VAL_MED=$i
@@ -82,18 +89,18 @@ if [[ "$CLASSIF" != "." ]] ; then
   if [[ -s "$PAR" ]] ; then
     for i in $(cat "$PAR" | awk "\$2==$VAL_CLS{print \$1}") ; do
       if [[ $METRIC == "aai" ]] ; then
-        AAI=$(noref_haai_or_aai $DATASET $i)
+        AAI=$(miga-noref_haai_or_aai $DATASET $i)
       else
         AAI=100
       fi
       if [[ $(perl -e "print 1 if '$AAI' >= 90") == "1" ]] ; then
-        noref_ani $DATASET $i
+        miga-noref_ani $DATASET $i
       fi
-      checkpoint_n
+      miga-checkpoint_n
     done
   fi
 fi
 
 #Finalize
 N=11
-checkpoint_n
+miga-checkpoint_n

data/scripts/_distances_ref_nomulti.bash

@@ -4,7 +4,7 @@
 
 set -e
 
-function checkpoint_n {
+fx_exists miga-checkpoint_n || function miga-checkpoint_n {
   if [[ $N -eq 10 ]] ; then
     for t in 01.haai 02.aai 03.ani ; do
       if [[ -s $TMPDIR/$t.db ]] ; then
@@ -36,41 +36,42 @@ for i in $(miga list_datasets -P "$PROJECT" --ref --no-multi) ; do
   # Check if the i-th dataset is ready
   [[ -s $ESS/$i.done && -s $ESS/$i.json ]] || continue
   # Check if this is done (e.g., in a previous failed iteration)
-  AAI=$(aai_from_db $DATASET $i $TMPDIR/02.aai.db)
+  AAI=$(miga-aai_from_db $DATASET $i $TMPDIR/02.aai.db)
   # Try the other direction
-  [[ "${AAI%.*}" -le 0 ]] && AAI=$(aai_from_db $i $DATASET 02.aai/$i.db)
+  [[ "${AAI%.*}" -le 0 ]] && AAI=$(miga-aai_from_db $i $DATASET 02.aai/$i.db)
   # Try with hAAI
   if [[ "${AAI%.*}" -le 0 ]] ; then
     [[ -e "$TMPDIR/$DATASET.ess.faa" ]] \
       || cp $ESS/$DATASET.ess.faa $TMPDIR/$DATASET.ess.faa
-    AAI=$(haai $TMPDIR/$DATASET.ess.faa $ESS/$i.ess.faa \
+    AAI=$(miga-haai $TMPDIR/$DATASET.ess.faa $ESS/$i.ess.faa \
       $CORES $TMPDIR/01.haai.db $TMPDIR/02.aai.db)
   fi
   # Try with complete AAI
   if [[ "${AAI%.*}" -le 0 ]] ; then
     [[ -e "$TMPDIR/$DATASET.faa" ]] \
       || cp ../06.cds/$DATASET.faa $TMPDIR/$DATASET.faa
-    AAI=$(aai $TMPDIR/$DATASET.faa ../06.cds/$i.faa $CORES $TMPDIR/02.aai.db)
+    AAI=$(miga-aai $TMPDIR/$DATASET.faa ../06.cds/$i.faa \
+      $CORES $TMPDIR/02.aai.db)
   fi
   # Check if ANI is meaningful
   if [[ -e "../05.assembly/$DATASET.LargeContigs.fna" \
       && -e "../05.assembly/$i.LargeContigs.fna" \
       && $(perl -e "print 1 if '$AAI' >= 90") == "1" ]] ; then
     # Check if this is done (e.g., in a previous failed iteration)
-    ANI=$(ani_from_db $DATASET $i $TMPDIR/03.ani.db)
+    ANI=$(miga-ani_from_db $DATASET $i $TMPDIR/03.ani.db)
     # Try the other direction
-    [[ "${ANI%.*}" -le 0 ]] && ANI=$(ani_from_db $i $DATASET 03.ani/$i.db)
+    [[ "${ANI%.*}" -le 0 ]] && ANI=$(miga-ani_from_db $i $DATASET 03.ani/$i.db)
     # Calculate it
     if [[ "${ANI%.*}" -le 0 ]] ; then
       [[ -e "$TMPDIR/$DATASET.LargeContigs.fna" ]] \
         || cp ../05.assembly/$DATASET.LargeContigs.fna \
           $TMPDIR/$DATASET.LargeContigs.fna
-      ANI=$(ani $TMPDIR/$DATASET.LargeContigs.fna \
+      ANI=$(miga-ani $TMPDIR/$DATASET.LargeContigs.fna \
         ../05.assembly/$i.LargeContigs.fna $CORES $TMPDIR/03.ani.db)
     fi
   fi
-  checkpoint_n
+  miga-checkpoint_n
 done
 N=10
-checkpoint_n
+miga-checkpoint_n
 

data/scripts/aai_distances.bash

@@ -1,6 +1,7 @@
 #!/bin/bash
 # Available variables: $PROJECT, $RUNTYPE, $MIGA, $CORES
 set -e
+SCRIPT="aai_distances"
 echo "MiGA: $MIGA"
 echo "Project: $PROJECT"
 source "$MIGA/scripts/miga.bash" || exit 1
@@ -15,21 +16,24 @@ DS=$(miga list_datasets -P "$PROJECT" --ref --no-multi)
 # Extract values
 echo "metric a b value sd n omega" | tr " " "\\t" >miga-project.txt
 for i in $DS ; do
-   echo "SELECT CASE WHEN omega!=0 THEN 'AAI' ELSE 'hAAI_AAI' END," \
-      " seq1, seq2, aai, sd, n, omega from aai;" \
-      | sqlite3 "$i.db" | tr "\\|" "\\t" >>miga-project.txt
-   echo "$i" >> miga-project.log
+  echo "SELECT CASE WHEN omega!=0 THEN 'AAI' ELSE 'hAAI_AAI' END," \
+    " seq1, seq2, aai, sd, n, omega from aai;" \
+    | sqlite3 "$i.db" | tr "\\|" "\\t" >>miga-project.txt
+  echo "$i" >> miga-project.log
 done
 
 # R-ify
 echo "
-aai <- read.table('miga-project.txt', sep='\\t', h=T);
+aai <- read.table('miga-project.txt', sep='\\t', h=T, as.is=TRUE);
 save(aai, file='miga-project.Rdata');
-h <- hist(aai[,'value'], breaks=100, plot=FALSE);
-write.table(
-   cbind(h[['breaks']][-length(h[['breaks']])],h[['breaks']][-1],h[['counts']]),
-   file='miga-project.hist', quote=FALSE, sep='\\t',
-   col.names=FALSE, row.names=FALSE);
+if(sum(aai[,'a'] != aai[,'b']) > 0){
+  h <- hist(aai[aai[,'a'] != aai[,'b'], 'value'], breaks=100, plot=FALSE);
+  write.table(
+    cbind(h[['breaks']][-length(h[['breaks']])],
+      h[['breaks']][-1], h[['counts']]),
+    file='miga-project.hist', quote=FALSE, sep='\\t',
+    col.names=FALSE, row.names=FALSE);
+}
 " | R --vanilla
 
 # Gzip
@@ -37,5 +41,4 @@ gzip -9 -f miga-project.txt
 
 # Finalize
 date "+%Y-%m-%d %H:%M:%S %z" > "miga-project.done"
-miga add_result -P "$PROJECT" -r aai_distances
-
+miga add_result -P "$PROJECT" -r "$SCRIPT"

data/scripts/ani_distances.bash

@@ -1,6 +1,7 @@
 #!/bin/bash
 # Available variables: $PROJECT, $RUNTYPE, $MIGA, $CORES
 set -e
+SCRIPT="ani_distances"
 echo "MiGA: $MIGA"
 echo "Project: $PROJECT"
 source "$MIGA/scripts/miga.bash" || exit 1
@@ -15,20 +16,23 @@ DS=$(miga list_datasets -P "$PROJECT" --ref --no-multi)
 # Extract values
 echo "metric a b value sd n omega" | tr " " "\\t" >miga-project.txt
 for i in $DS ; do
-   echo "SELECT 'ANI', seq1, seq2, ani, sd, n, omega from ani ;" \
-      | sqlite3 "$i.db" | tr "\\|" "\\t" >>miga-project.txt
-   echo "$i" >> miga-project.log
+  echo "SELECT 'ANI', seq1, seq2, ani, sd, n, omega from ani ;" \
+    | sqlite3 "$i.db" | tr "\\|" "\\t" >>miga-project.txt
+  echo "$i" >> miga-project.log
 done
 
 # R-ify
 echo "
-ani <- read.table('miga-project.txt', sep='\\t', h=T);
+ani <- read.table('miga-project.txt', sep='\\t', h=T, as.is=TRUE);
 save(ani, file='miga-project.Rdata');
-h <- hist(ani[,'value'], breaks=100, plot=FALSE);
-write.table(
-   cbind(h[['breaks']][-length(h[['breaks']])],h[['breaks']][-1],h[['counts']]),
-   file='miga-project.hist', quote=FALSE, sep='\\t',
-   col.names=FALSE, row.names=FALSE);
+if(sum(ani[,'a'] != ani[,'b']) > 0){
+  h <- hist(ani[ani[,'a'] != ani[,'b'], 'value'], breaks=100, plot=FALSE);
+  write.table(
+    cbind(h[['breaks']][-length(h[['breaks']])],
+      h[['breaks']][-1], h[['counts']]),
+    file='miga-project.hist', quote=FALSE, sep='\\t',
+    col.names=FALSE, row.names=FALSE);
+}
 " | R --vanilla
 
 # Gzip
@@ -36,5 +40,4 @@ gzip -9 -f miga-project.txt
 
 # Finalize
 date "+%Y-%m-%d %H:%M:%S %z" > "miga-project.done"
-miga add_result -P "$PROJECT" -r ani_distances
-
+miga add_result -P "$PROJECT" -r "$SCRIPT"

data/scripts/assembly.bash

@@ -1,6 +1,7 @@
 #!/bin/bash
 # Available variables: $PROJECT, $RUNTYPE, $MIGA, $CORES
 set -e
+SCRIPT="assembly"
 echo "MiGA: $MIGA"
 echo "Project: $PROJECT"
 source "$MIGA/scripts/miga.bash" || exit 1
@@ -49,4 +50,4 @@ FastA.length.pl $DATASET.AllContigs.fna | awk '$2>=1000{print $1}' \
 
 # Finalize
 date "+%Y-%m-%d %H:%M:%S %z" > "$DATASET.done"
-miga add_result -P "$PROJECT" -D "$DATASET" -r assembly
+miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT"

data/scripts/cds.bash

@@ -1,6 +1,7 @@
 #!/bin/bash
 # Available variables: $PROJECT, $RUNTYPE, $MIGA, $CORES
 set -e
+SCRIPT="cds"
 echo "MiGA: $MIGA"
 echo "Project: $PROJECT"
 source "$MIGA/scripts/miga.bash" || exit 1
@@ -42,5 +43,4 @@ gzip -9 -f "$DATASET.gff2"
 
 # Finalize
 date "+%Y-%m-%d %H:%M:%S %z" > "$DATASET.done"
-miga add_result -P "$PROJECT" -D "$DATASET" -r cds
-
+miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT"

data/scripts/clade_finding.bash

@@ -1,6 +1,7 @@
 #!/bin/bash
 # Available variables: $PROJECT, $RUNTYPE, $MIGA, $CORES
 set -e
+SCRIPT="clade_finding"
 echo "MiGA: $MIGA"
 echo "Project: $PROJECT"
 source "$MIGA/scripts/miga.bash" || exit 1
@@ -42,4 +43,4 @@ fi
 
 # Finalize
 date "+%Y-%m-%d %H:%M:%S %z" > "miga-project.done"
-miga add_result -P "$PROJECT" -r clade_finding
+miga add_result -P "$PROJECT" -r "$SCRIPT"

data/scripts/distances.bash

@@ -1,6 +1,7 @@
 #!/bin/bash
 # Available variables: $PROJECT, $DATASET, $RUNTYPE, $MIGA, $CORES
 set -e
+SCRIPT="distances"
 echo "MiGA: $MIGA"
 echo "Project: $PROJECT"
 source "$MIGA/scripts/miga.bash" || exit 1
@@ -28,5 +29,4 @@ fi
 # Finalize
 rm -R $TMPDIR
 date "+%Y-%m-%d %H:%M:%S %z" > "$DATASET.done"
-miga add_result -P "$PROJECT" -D "$DATASET" -r distances
-
+miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT"