Versions are now captured in the result JSONs

Towards #143

lib/miga/cli/action/add_result.rb

@@ -5,13 +5,17 @@
 
 class MiGA::Cli::Action::AddResult < MiGA::Cli::Action
   def parse_cli
-    cli.defaults = { force: false }
+    cli.defaults = { force: false, stdin_versions: 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 }
+      opt.on(
+        '--stdin-versions',
+        'Read Software versions from STDIN'
+      ) { |v| cli[:stdin_versions] = v }
     end
   end
 
@@ -21,5 +25,22 @@ def perform
     cli.say "Registering result: #{cli[:result]}"
     r = obj.add_result(cli[:result], true, force: cli[:force])
     raise 'Cannot add result, incomplete expected files' if r.nil?
+
+    # Add Software version data
+    if cli[:stdin_versions]
+      versions = {}
+      sw = nil
+      $stdin.each do |ln|
+        ln = ln.chomp.strip
+        if ln =~ /^=> (.*)/
+          sw = $1
+          versions[sw] = ''
+        else
+          versions[sw] += ln
+        end
+      end
+      r.add_versions(versions)
+      r.save
+    end
   end
 end

lib/miga/result.rb

@@ -3,13 +3,15 @@
 require 'miga/result/dates'
 require 'miga/result/source'
 require 'miga/result/stats'
+require 'miga/result/versions'
 
 ##
 # The result from a task run. It can be project-wide or dataset-specific.
 class MiGA::Result < MiGA::MiGA
   include MiGA::Result::Dates
   include MiGA::Result::Source
   include MiGA::Result::Stats
+  include MiGA::Result::Versions
 
   # Class-level
   class << self
@@ -151,7 +153,11 @@ def add_files(files)
   ##
   # Initialize and #save empty result
   def create
-    @data = { created: Time.now.to_s, stats: {}, files: {} }
+    @data = {
+      created: Time.now.to_s,
+      stats: {}, files: {},
+      versions: { 'MiGA' => MiGA::VERSION.join('.') }
+    }
     save
   end
 

lib/miga/result/versions.rb

@@ -0,0 +1,17 @@
+require 'miga/result/base'
+
+##
+# Helper module including functions for results to handle software versions
+module MiGA::Result::Versions
+  ##
+  # Add version information for the Software used by this result
+  def add_versions(versions)
+    versions.each { |k, v| self[:versions][k] = v }
+  end
+
+  ##
+  # Get list of software and their versions as raw text (Markdown)
+  def versions_md
+    versions.map { |k, v| "- #{k}: #{v}" }.join("\n")
+  end
+end

scripts/assembly.bash

@@ -10,6 +10,7 @@ cd "$PROJECT/data/05.assembly"
 miga date > "$DATASET.start"
 
 # Interpose (if needed)
+interpose=no
 TF="../04.trimmed_fasta"
 b=$DATASET
 if [[ -s "$TF/${b}.2.fasta" || -s "$TF/${b}.2.fasta.gz" ]] ; then
@@ -22,6 +23,7 @@ if [[ -s "$TF/${b}.2.fasta" || -s "$TF/${b}.2.fasta.gz" ]] ; then
         gzip -cd "$TF/${b}.${s}.fasta.gz" > "${b}.${s}.tmp"
       fi
     done
+    interpose=yes
     FastA.interpose.pl "$cr" "$b".[12].tmp
     rm "$b".[12].tmp
     miga add_result -P "$PROJECT" -D "$DATASET" -r trimmed_fasta -f
@@ -62,5 +64,17 @@ FastA.length.pl "$DATASET.AllContigs.fna" | awk '$2>=1000{print $1}' \
 
 # Finalize
 miga date > "$DATASET.done"
-miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f
+cat <<VERSIONS \
+  | miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f --stdin-versions
+=> MiGA
+$(miga --version)
+$(
+  if [[ "$interpose" == "yes" ]] ; then
+    echo "=> Enveomics Collection: FastA.interpose.pl"
+    echo "version unknown"
+  fi
+)
+=> IDBA-UD
+version unknown
+VERSIONS
 

scripts/cds.bash

@@ -68,6 +68,12 @@ for ext in gff3 faa fna ; do
 done
 
 # Finalize
-miga date > "$DATASET.done"
-miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f
+miga date > "${DATASET}.done"
+cat <<VERSIONS \
+  | miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f --stdin-versions
+=> MiGA
+$(miga --version)
+=> Prodigal
+$(prodigal -v 2>&1 | grep . | perl -pe 's/^Prodigal //')
+VERSIONS
 

scripts/distances.bash

@@ -22,5 +22,99 @@ fi
 ruby -I "$MIGA/lib" "$MIGA/utils/distances.rb" "$PROJECT" "$DATASET"
 
 # Finalize
-miga date > "$DATASET.done"
-miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f
+fastaai=no
+aai=no
+ani=no
+blast=no
+blat=no
+diamond=no
+fastani=no
+case $(miga option -P "$PROJECT" -k haai_p) in
+  fastaai)
+    fastaai=yes
+    ;;
+  diamond)
+    diamond=yes
+    aai=yes
+    ;;
+  blast)
+    blast=yes
+    aai=yes
+    ;;
+esac
+
+case $(miga option -P "$PROJECT" -k aai_p) in
+  diamond)
+    diamond=yes
+    aai=yes
+    ;;
+  blast)
+    blast=yes
+    aai=yes
+    ;;
+esac
+
+case $(miga option -P "$PROJECT" -k ani_p) in
+  blast)
+    blast=yes
+    ani=yes
+    ;;
+  blat)
+    blat=yes
+    ani=yes
+    ;;
+  fastani)
+    fastani=yes
+    ;;
+esac
+
+
+miga date > "${DATASET}.done"
+cat <<VERSIONS \
+  | miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f --stdin-versions
+=> MiGA
+$(miga --version)
+$(
+  if [[ "$fastaai" == "yes" ]] ; then
+    echo "=> FastAAI"
+    fastaai version 2>&1 | perl -pe 's/.*=//'
+  fi
+)
+$(
+  if [[ "$fastani" == "yes" ]] ; then
+    echo "=> FastANI"
+    fastANI --version 2>&1 | grep . | perl -pe 's/^version //'
+  fi
+)
+$(
+  if [[ "$aai" == "yes" ]] ; then
+    echo "=> Enveomics Collection: aai.rb"
+    aai.rb --version 2>&1 | perl -pe 's/.*: //'
+  fi
+)
+$(
+  if [[ "$ani" == "yes" ]] ; then
+    echo "=> Enveomics Collection: ani.rb"
+    ani.rb --version 2>&1 | perl -pe 's/.*: //'
+  fi
+)
+$(
+  if [[ "$blast" == "yes" ]] ; then
+    echo "=> NCBI BLAST+"
+    blastp -version 2>&1 | tail -n 1 | perl -pe 's/.*: blast //'
+  fi
+)
+$(
+  if [[ "$blat" == "yes" ]] ; then
+    echo "=> BLAT"
+    blat 2>&1 | head -n 1 | perl -pe 's/.* v\. //' | perl -pe 's/ fast .*//'
+  fi
+)
+$(
+  if [[ "$diamond" == "yes" ]] ; then
+    echo "=> Diamond"
+    diamond --version 2>&1 | perl -pe 's/^diamond version //'
+  fi
+)
+VERSIONS
+

scripts/essential_genes.bash

@@ -70,4 +70,17 @@ fi
 
 # Finalize
 miga date > "${DATASET}.done"
-miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f
+cat <<VERSIONS \
+  | miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f --stdin-versions
+=> MiGA
+$(miga --version)
+=> Enveomics Collection: HMM.essential.rb
+$(HMM.essential.rb --version 2>&1 | perl -pe 's/.*: //')
+$(
+  if [[ "$NOMULTI" -eq "1" ]] ; then
+    echo "=> FastAAI"
+    fastaai version 2>&1 | perl -pe 's/.*=//'
+  fi
+)
+VERSIONS
+

scripts/mytaxa.bash

@@ -98,5 +98,20 @@ else
 fi
 
 # Finalize
-miga date > "$DATASET.done"
-miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f
+miga date > "${DATASET}.done"
+cat <<VERSIONS \
+  | miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f --stdin-versions
+=> MiGA
+$(miga --version)
+$(
+  if [[ "$MIGA_MYTAXA" != "no" && "$MULTI" -eq "1" ]] ; then
+    echo "=> MyTaxa"
+    MyTaxa | grep Version: | perl -pe 's/.*: //'
+    echo "=> Diamond"
+    diamond --version 2>&1 | perl -pe 's/^diamond version //'
+    echo "=> Krona"
+    ktImportText | head -n 2 | tail -n 1 | awk '{ print $3 }'
+  fi
+)
+VERSIONS
+

scripts/mytaxa_scan.bash

@@ -97,5 +97,18 @@ else
 fi
 
 # Finalize
-miga date > "$DATASET.done"
-miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f
+miga date > "${DATASET}.done"
+cat <<VERSIONS \
+  | miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f --stdin-versions
+=> MiGA
+$(miga --version)
+$(
+  if [[ "$MIGA_MYTAXA" != "no" && "$NOMULTI" -eq "1" ]] ; then
+    echo "=> MyTaxa"
+    MyTaxa | grep Version: | perl -pe 's/.*: //'
+    echo "=> Diamond"
+    diamond --version 2>&1 | perl -pe 's/^diamond version //'
+  fi
+)
+VERSIONS
+

scripts/read_quality.bash

@@ -19,6 +19,10 @@ for s in 1 2 ; do
 done
 
 # Finalize
-miga date > "$DATASET.done"
-miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f
+miga date > "${DATASET}.done"
+cat <<VERSIONS \
+  | miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f --stdin-versions
+=> MiGA
+$(miga --version)
+VERSIONS
 

scripts/ssu.bash

@@ -65,4 +65,22 @@ fi
 
 # Finalize
 miga date > "${DATASET}.done"
-miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f
+cat <<VERSIONS \
+  | miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f --stdin-versions
+=> MiGA
+$(miga --version)
+$(
+  if [[ -s $fa ]] ; then
+    echo "=> barrnap"
+    barrnap --version 2>&1 | perl -pe 's/^barrnap //'
+    echo "=> bedtools"
+    bedtools --version 2>&1 | perl -pe 's/^bedtools //'
+    echo "=> Enveomics Collection"
+    echo "version unknown"
+    echo "=> RDP Naive Bayes Classifier"
+    tail -n 1 "${DATASET}.rdp.tsv" | perl -pe 's/.*: //'
+    echo "=> tRNAscan-SE"
+    tRNAscan-SE -h 2>&1 | head -n 2 | tail -n 1 | perl -pe 's/^tRNAscan-SE //'
+  fi
+)
+VERSIONS

scripts/stats.bash

@@ -12,11 +12,17 @@ cd "$DIR"
 miga date > "${DATASET}.start"
 
 # Calculate statistics
-for i in raw_reads trimmed_fasta assembly cds essential_genes distances taxonomy ssu ; do
+for i in raw_reads trimmed_fasta assembly \
+         cds essential_genes distances taxonomy ssu ; do
   echo "# $i"
   miga stats --compute-and-save --ignore-empty -P "$PROJECT" -D "$DATASET" -r $i
 done
 
 # Finalize
-miga date > "$DATASET.done"
-miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f
+miga date > "${DATASET}.done"
+cat <<VERSIONS \
+  | miga add_result -P "$PROJECT" -D "$DATASET" -r "$SCRIPT" -f --stdin-versions
+=> MiGA
+$(miga --version)
+VERSIONS
+