miga-base 1.0.5.2 → 1.1.0.0

data/utils/FastAAI/{FastAAI → FastAAI-legacy}/FastAAI

@@ -226,7 +226,7 @@ def kmer_extract(input_files):
             line = line.strip().split()
             protein_name = line[0]
             model_name = line[3]
-            score = line[8]
+            score = float(line[8])
             if model_name in positive_matches:
                 if score > positive_matches[model_name][1]:
                     positive_matches[model_name] = [protein_name, score]

data/utils/distance/commands.rb

@@ -141,32 +141,43 @@ module MiGA::DistanceRunner::Commands
   ##
   # Execute a FastAAI command
   def fastaai_cmd(targets)
-    qry_idx = dataset.result(:essential_genes).file_path(:fastaai_index)
+    qry_idx = dataset.result(:essential_genes).file_path(:fastaai_index_2)
     return nil unless qry_idx
 
-    # Run FastAAI
-    File.open(f1 = tmp_file, 'w') { |fh| fh.puts qry_idx }
-    File.open(f2 = tmp_file, 'w') do |fh|
-      targets.each do |target|
-        target_idx = target&.result(:essential_genes)&.file_path(:fastaai_index)
-        fh.puts target_idx if target_idx
-      end
+    # Merge databases
+    donors = []
+    targets.each do |target|
+      tgt_idx = target&.result(:essential_genes)&.file_path(:fastaai_index_2)
+      donors << tgt_idx if tgt_idx
     end
+    return nil if donors.empty?
     run_cmd <<~CMD
-              FastAAI --qd "#{f1}" --rd "#{f2}" --output "#{f3 = tmp_file}" \
-              --threads #{opts[:thr]}
+              FastAAI merge_db --donors "#{donors.join(',')}" \
+                --recipient "#{f1 = tmp_file}" --threads #{opts[:thr]}
+            CMD
+
+    # Run FastAAI
+    run_cmd <<~CMD
+              FastAAI db_query --query "#{qry_idx}" --target "#{f1}" \
+                --output "#{f2 = tmp_file}" --threads #{opts[:thr]} \
+                --do_stdev
             CMD
 
     # Save values in the databases
     haai_data = {}
     aai_data = {}
-    File.open(f3, 'r') do |fh|
+    # Ugly workaround to the insistence of FastAAI to not provide the files
+    # I ask for ;-)
+    qry_results = File.basename(qry_idx, '.faix') + '_results.txt'
+    out_file = File.join(f2, 'results', qry_results)
+    File.open(out_file, 'r') do |fh|
       fh.each do |ln|
         out = ln.chomp.split("\t")
         haai_data[out[1]] = [
           out[2].to_f * 100, out[3].to_f * 100, out[4].to_i, out[5].to_i
         ]
-        aai_data[out[1]] = [out[6].to_f, 0, 0, 0] if out[6] !~ /^>/
+        out[6] = (out[6] =~ /^>/) ? 95.0 : out[6].to_f
+        aai_data[out[1]] = [out[6], 0, 0, 0]
       end
     end
     puts "Results: #{haai_data.size} | Inferences: #{aai_data.size}"
@@ -174,7 +185,7 @@ module MiGA::DistanceRunner::Commands
     batch_data_to_db(:aai, aai_data)
 
     # Cleanup
-    [f1, f2, f3].each { |i| File.unlink(i) }
+    [f1, f2].each { |i| FileUtils.rm_r(i) }
   end
 
   ##

data/utils/requirements.txt

@@ -1,12 +1,12 @@
-Software	Test exec	Website								Notes
---------	---------	-------								-----
-Ruby		ruby		https://www.ruby-lang.org/					Required version: 2.3+
-Python		python3		https://www.python.org/						Required version: 3+
+Software	Test exec	Website						Notes
+--------	---------	-------						-----
+Ruby		ruby		https://www.ruby-lang.org/			Required version: 2.3+
+Python		python3		https://www.python.org/				Required version: 3+
 R		R		http://www.r-project.org/
 SQLite3		sqlite3		https://www.sqlite.org/
 NCBI BLAST+	blastp		ftp://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/LATEST
-DIAMOND		diamond		http://ab.inf.uni-tuebingen.de/software/diamond			Required version: v0.9.20+
-FastANI		fastANI		https://github.com/ParBLiSS/FastANI				Required version: 1.1+
+DIAMOND		diamond		http://ab.inf.uni-tuebingen.de/software/diamond	Required version: v0.9.20+
+FastANI		fastANI		https://github.com/ParBLiSS/FastANI		Required version: 1.1+
 HMMer 3.0+	hmmsearch	http://hmmer.janelia.org/software
 Bedtools	bedtools	http://bedtools.readthedocs.org/en/latest/
 Prodigal	prodigal	http://prodigal.ornl.gov
@@ -17,6 +17,6 @@ FaQCs (reads)	FaQCs		https://github.com/LANL-Bioinformatics/FaQCs
 Falco (reads)	falco		https://github.com/smithlabcode/falco
 Seqtk (reads)	seqtk		https://github.com/lh3/seqtk
 Fastp (reads)	fastp		https://github.com/OpenGene/fastp
-OpenJDK (rdp)	java		https://adoptopenjdk.net/					Any Java VM would work
+Temurin (rdp)	java		https://adoptium.net/				Any Java VM would work
 MyTaxa (mytaxa)	MyTaxa		http://enve-omics.ce.gatech.edu/mytaxa
 Krona (mytaxa)	ktImportText	https://github.com/marbl/Krona/wiki

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: miga-base
 version: !ruby/object:Gem::Version
-  version: 1.0.5.2
+  version: 1.1.0.0
 platform: ruby
 authors:
 - Luis M. Rodriguez-R
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2021-08-26 00:00:00.000000000 Z
+date: 2021-10-28 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: daemons
@@ -107,6 +107,7 @@ files:
 - README.md
 - Rakefile
 - bin/miga
+- bin/miga-env
 - lib/miga.rb
 - lib/miga/_data/aai-intax-blast.tsv.gz
 - lib/miga/_data/aai-intax-diamond.tsv.gz
@@ -155,6 +156,7 @@ files:
 - lib/miga/cli/action/ln.rb
 - lib/miga/cli/action/ls.rb
 - lib/miga/cli/action/ncbi_get.rb
+- lib/miga/cli/action/ncbi_get/downloads.rb
 - lib/miga/cli/action/new.rb
 - lib/miga/cli/action/next_step.rb
 - lib/miga/cli/action/option.rb
@@ -260,9 +262,10 @@ files:
 - utils/FastAAI/00.Libraries/01.SCG_HMMs/Archaea_SCG.hmm
 - utils/FastAAI/00.Libraries/01.SCG_HMMs/Bacteria_SCG.hmm
 - utils/FastAAI/00.Libraries/01.SCG_HMMs/Complete_SCG_DB.hmm
-- utils/FastAAI/FastAAI/FastAAI
+- utils/FastAAI/FastAAI
+- utils/FastAAI/FastAAI-legacy/FastAAI
+- utils/FastAAI/FastAAI-legacy/kAAI_v1.0_virus.py
 - utils/FastAAI/README.md
-- utils/FastAAI/kAAI_v1.0_virus.py
 - utils/adapters.fa
 - utils/cleanup-databases.rb
 - utils/core-pan-plot.R