rbbt-util 5.37.8 → 5.37.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b5c4e63f52c80871ada12e77c2c2cfe0d2cacb9d4bb3cf13fde88f814d30987b
-  data.tar.gz: e3859c1668e92be90edf4ca5637480e8d74aa26822c0f3181f8d9a1689e6dc0b
+  metadata.gz: 2f5b8f1cf0bc5999c1b467eb794cf337e0a315ed24ca3de3809a7d9eff73e175
+  data.tar.gz: 3d2f6a70bba07a441cc6ae7325e0ecd149440173975e8b6ba7e3328d858795a5
 SHA512:
-  metadata.gz: 1f15f9954cf3f4ad93859d98155c6b34af5beeba3286367732a71c2e9093d97477ec405d300eded639e93470bf8b8f466014e79eaaddedb621c0df2862ac0e01
-  data.tar.gz: 80f82e5ca9262aad1ce8300de0a486234017c6a3a450ae9579bbbeeca1062dac4ab42d3c31276d5546d882a3efa32520658efec18847099c0ae5f8ee93daab7b
+  metadata.gz: a02576707e38808f545af7d7a17556b9e38ac84ccb203d320ce3ed7d7dd6b5f1c0834c2b26e726e68b8b731d517de4ce2694956c86c565c19329380ffee1a037
+  data.tar.gz: 71813abce39b34076f97e16b8d2c26ff8e2f3287a75cf43e15cd3ed97205c4e4d8239e19b4fa670d0129e6de81c42bb1f6b1c4ec908f3554e6ac54c56106bb66

data/lib/rbbt/knowledge_base/enrichment.rb

@@ -1,7 +1,7 @@
 require 'rbbt/knowledge_base/registry'
-require 'rbbt/statistics/hypergeometric'
 class KnowledgeBase
   def enrichment(name, entities, options = {})
+    require 'rbbt/statistics/hypergeometric'
     database = get_database(name, options)
     entities = identify_source name, entities
     database.enrichment entities, database.fields.first, :persist => false

data/lib/rbbt/resource/util.rb

@@ -1,6 +1,6 @@
 module Path
 
-  def self.caller_lib_dir(file = nil, relative_to = ['lib', 'bin'])
+  def self.caller_lib_dir(file = nil, relative_to = ['lib', 'bin', 'workflow.rb'])
     #file = caller.reject{|l| 
     #  l =~ /rbbt\/(?:resource\.rb|workflow\.rb)/ or
     #  l =~ /rbbt\/resource\/path\.rb/ or

data/lib/rbbt/util/migrate.rb

@@ -61,7 +61,7 @@ puts resource[path].find(search_path)
 
     real_paths.each do |source_path|
       Log.medium "Migrating #{source_path} #{options[:files].length} files to #{target} - #{Misc.fingerprint(options[:files])}}" if options[:files]
-      if File.directory?(source_path) || source_path.ends_with?("/")
+      if File.directory?(source_path) || source_path.end_with?("/")
         source_path += "/" unless source_path.end_with? '/'
         target += "/" unless target.end_with? '/'
       end

data/lib/rbbt/workflow/examples.rb

@@ -25,27 +25,32 @@ module Workflow
     end.compact
   end
 
+  #def example_inputs_old(task_name, example)
+  #  inputs = {}
+  #  IndiferentHash.setup(inputs)
+  #  example(task_name, example).each do |input,type,file|
+
+  #    case type
+  #    when :tsv, :array, :text
+  #      Log.debug "Pointing #{ input } to #{file}"
+  #      inputs[input.to_sym]  = file
+  #    when :boolean
+  #      inputs[input.to_sym]  = (file.read.strip == 'true')
+  #    else
+  #      Log.debug "Loading #{ input } from #{file}"
+  #      inputs[input.to_sym]  = file.read.strip
+  #    end
+  #  end
+  #  inputs
+  #end
+
   def example_inputs(task_name, example)
-    inputs = {}
-    IndiferentHash.setup(inputs)
-    example(task_name, example).each do |input,type,file|
-
-      case type
-      when :tsv, :array, :text
-        Log.debug "Pointing #{ input } to #{file}"
-        inputs[input.to_sym]  = file
-      when :boolean
-        inputs[input.to_sym]  = (file.read.strip == 'true')
-      else
-        Log.debug "Loading #{ input } from #{file}"
-        inputs[input.to_sym]  = file.read.strip
-      end
-    end
-    inputs
+    dir = example_dir[task_name][example]
+    info = self.task_info(task_name)
+    Workflow.load_inputs(dir, info[:inputs], info[:input_types])
   end
 
-
-  def example_step(task_name, example, new_inputs = {})
+  def example_step(task_name, example="Example", new_inputs = {})
     inputs = example_inputs(task_name, example)
 
     if new_inputs and new_inputs.any?

data/lib/rbbt/workflow/step/save_load_inputs.rb

@@ -250,59 +250,4 @@ class Step
     inputs.keys
   end
 
-  #def self.save_inputs_old(inputs, input_types, input_options, dir)
-  #  inputs.each do |name,value|
-  #    type = input_types[name]
-  #    type = type.to_s if type
-  #    path = File.join(dir, name.to_s)
-
-  #    if (IO === value || Step === value) && ! (input_options[name] && input_options[name][:nofile])
-  #      path = path + '.as_io' 
-  #    elsif Misc.is_filename?(value, false)
-  #      path = path + '.as_filename'
-  #    end
-
-  #    Log.debug "Saving job input #{name} (#{type}) into #{path}"
-
-  #    case
-  #    when IO === value
-  #      Open.write(path, value.to_s)
-  #    when Step === value
-  #      Open.ln_s(value.path, path)
-  #    when type.to_s == "binary"
-  #      if String === value && File.exist?(value)
-  #        value = File.expand_path(value)
-  #        Open.ln_s(value, path)
-  #      elsif String === value && Misc.is_filename?(value, false)
-  #        Open.write(path + '.as_path' , value)
-  #      else
-  #        Open.write(path, value, :mode => 'wb')
-  #      end
-  #    when TSV === value
-  #      Open.write(path, value.to_s)
-  #    when Array === value
-  #      Open.write(path, value.collect{|v| Step === v ? v.path : v.to_s} * "\n")
-  #    when %w(file tsv array).include?(type.to_s)
-  #      if String === value && File.exist?(value)
-  #        value = File.expand_path(value)
-  #        Open.ln_s(value, path)
-  #      elsif String === value && Misc.is_filename?(value, false)
-  #        Open.write(path + '.as_path' , value)
-  #      else
-  #        value = value.collect{|v| v = "#{v}" if Path === v; v } if Array === value
-  #        value = "#{value}" if Path === value
-  #        Open.write(path + '.yaml', value.to_yaml)
-  #      end
-  #    when IO === value
-  #      if value.filename && String === value.filename && File.exist?(value.filename)
-  #        Open.ln_s(value.filename, path)
-  #      else
-  #        Open.write(path, value)
-  #      end
-  #    else
-  #      Open.write(path, value.to_s)
-  #    end
-  #  end.any?
-  #end
-
 end

data/lib/rbbt/workflow.rb

@@ -206,7 +206,7 @@ module Workflow
         (Workflow.autoinstall and `rbbt workflow install #{Misc.snake_case(wf_name)} || rbbt workflow install #{wf_name}` and require_local_workflow(wf_name)) or raise("Workflow not found or could not be loaded: #{ wf_name }")
 
       workflow = begin
-                   Misc.string2const Misc.camel_case(wf_name.split("+").first)
+                   Misc.string2const Misc.camel_case(wf_name)
                  rescue
                    Workflow.workflows.last || true
                  end
@@ -370,11 +370,15 @@ module Workflow
   def exec_exports
     @exec_exports ||= []
   end
-  
+
   def all_exports
     @all_exports ||= asynchronous_exports + synchronous_exports + exec_exports + stream_exports
   end
 
+  def documented_tasks
+    tasks.select{|k,v| v.description && ! v.description.empty? }.collect{|k,v| k }
+  end
+
   # {{{ JOB MANAGEMENT
   DEFAULT_NAME="Default"
 

data/share/rbbt_commands/app/start

@@ -2,6 +2,8 @@
 
 require 'rbbt-util'
 require 'rbbt/util/simpleopt'
+require 'rack'
+require 'rack/server'
 
 $0 = "rbbt #{$previous_commands*""} #{ File.basename(__FILE__) }" if $previous_commands
 
@@ -60,7 +62,7 @@ app_dir = app_dir[app].find
 
 server = options[:server] || 'puma'
 Misc.in_dir(app_dir) do
-  require 'rack'
+
   ENV["RBBT_FINDER"] = "true" if options[:finder]
   ENV["RACK_ENV"] = options[:environment] if options.include?(:environment)
   ENV["RBBT_VIEWS_DIR"] = options[:views] if options.include?(:views)

data/test/rbbt/association/test_database.rb

@@ -6,50 +6,59 @@ require 'rbbt/association/database'
 
 class TestAssociationDatabase < Test::Unit::TestCase
    
-  EFFECT =<<-END
+  def effect_options
+    effect_options = {
+      :source => "SG=~Associated Gene Name",
+      :target => "TG=~Associated Gene Name=>Ensembl Gene ID",
+      :persist => false,
+    :identifiers => datafile_test('identifiers'),
+    }
+  end
+
+  def effect_tsv
+    effect =<<-END
 #: :sep=" "#:type=:double
 #SG TG Effect directed?
 MDM2 TP53 inhibition false
 TP53 NFKB1|GLI1 activation|activation true|true
-  END
+    END
 
-  EFFECT_OPTIONS = {
-    :source => "SG=~Associated Gene Name",
-    :target => "TG=~Associated Gene Name=>Ensembl Gene ID",
-    :persist => false,
-    :identifiers => datafile_test('identifiers'),
-  }
 
-  EFFECT_TSV = TSV.open EFFECT, EFFECT_OPTIONS.dup
+    effect_tsv = TSV.open effect, effect_options.dup
+  end
+
+  def effect_database
+    effect_database = Association.database(effect_tsv)
+  end
+
   def test_tsv
-    assert_equal %w(NFKB1 GLI1), EFFECT_TSV["TP53"]["TG"]
+    assert_equal %w(NFKB1 GLI1), effect_tsv["TP53"]["TG"]
   end
 
-  EFFECT_DATABASE = Association.database(EFFECT_TSV, EFFECT_OPTIONS.dup)
   
   def test_open_new_tsv
-    tsv = Association.database EFFECT_TSV, :fields => ["Effect"], :target => "TG=~Associated Gene Name=>Ensembl Gene ID"
+    tsv = Association.database effect_tsv, :fields => ["Effect"], :target => "TG=~Associated Gene Name=>Ensembl Gene ID"
     assert_equal ["Ensembl Gene ID", "Effect"], tsv.fields
     assert_equal "SG", tsv.key_field
     assert tsv.include? "MDM2"
   end
 
   def test_open_new_tsv_reverse
-    tsv = Association.database EFFECT_TSV, :fields => ["Effect"], :target => "SG=~Associated Gene Name=>Ensembl Gene ID", :zipped => true
+    tsv = Association.database effect_tsv, :fields => ["Effect"], :target => "SG=~Associated Gene Name=>Ensembl Gene ID", :zipped => true
     assert_equal ["Ensembl Gene ID", "Effect"], tsv.fields
     assert_equal "TG", tsv.key_field
     assert tsv.include? "NFKB1"
   end
 
   def test_database_translate
-    database = Association.database(EFFECT_TSV, EFFECT_OPTIONS.merge({:source => "SG=~Associated Gene Name=>Ensembl Gene ID", :target => "TG"}))
+    database = Association.database(effect_tsv, effect_options.merge({:source => "SG=~Associated Gene Name=>Ensembl Gene ID", :target => "TG"}))
     assert_equal %w(inhibition), database["ENSG00000135679"]["Effect"]
     assert_equal %w(activation activation), database["ENSG00000141510"]["Effect"]
   end
 
   def test_database
-    assert_equal %w(inhibition), EFFECT_DATABASE["MDM2"]["Effect"]
-    assert_equal %w(activation activation), EFFECT_DATABASE["TP53"]["Effect"]
+    assert_equal %w(inhibition), effect_database["MDM2"]["Effect"]
+    assert_equal %w(activation activation), effect_database["TP53"]["Effect"]
   end
 
   def test_index_list

data/test/rbbt/entity/test_identifiers.rb

@@ -7,14 +7,16 @@ module Gene
   extend Entity
 end
 
-require 'rbbt/sources/kegg'
-require 'rbbt/sources/organism'
-Gene.add_identifiers Organism.identifiers("NAMESPACE"), "Ensembl Gene ID", "Associated Gene Name"
-Gene.add_identifiers KEGG.identifiers
-Gene.add_identifiers Organism.identifiers("NAMESPACE"), "Ensembl Gene ID", "Associated Gene Name"
-Gene.add_identifiers KEGG.identifiers
-
 class TestEntityIdentifiers < Test::Unit::TestCase
+  def setup
+    require 'rbbt/sources/kegg'
+    require 'rbbt/sources/organism'
+    Gene.add_identifiers Organism.identifiers("NAMESPACE"), "Ensembl Gene ID", "Associated Gene Name"
+    Gene.add_identifiers KEGG.identifiers
+    Gene.add_identifiers Organism.identifiers("NAMESPACE"), "Ensembl Gene ID", "Associated Gene Name"
+    Gene.add_identifiers KEGG.identifiers
+
+  end
   def test_name
     Gene.add_identifiers datafile_test('identifiers'), "Ensembl Gene ID", "Associated Gene Name"
     assert_equal "TP53", Gene.setup("ENSG00000141510").name
@@ -62,7 +64,7 @@ class TestEntityIdentifiers < Test::Unit::TestCase
     mod.instance_eval do
       extend Entity
       add_identifiers Path.setup(description_file), key_field, description_field
-      
+
       annotation :format
     end
     entity = "CORUM:6052"

data/test/rbbt/hpc/test_batch.rb

@@ -2,10 +2,13 @@ require File.join(File.expand_path(File.dirname(__FILE__)), '../..', 'test_helpe
 require 'rbbt/hpc/batch'
 require 'rbbt/workflow'
 
-Workflow.require_workflow "Sample"
-Workflow.require_workflow "HTS"
 class TestSLURM < Test::Unit::TestCase
 
+  def setup
+    Workflow.require_workflow "Sample"
+    Workflow.require_workflow "HTS"
+  end
+
   def test_batch_options
     job = Sample.job(:mutect2, "small", :reference => "hg38")
 

data/test/rbbt/hpc/test_slurm.rb

@@ -2,11 +2,13 @@ require File.join(File.expand_path(File.dirname(__FILE__)), '../..', 'test_helpe
 require 'rbbt/hpc/slurm'
 require 'rbbt/workflow'
 
-Workflow.require_workflow "Sample"
-Workflow.require_workflow "HTS"
 class TestSLURM < Test::Unit::TestCase
+  def setup
+    Workflow.require_workflow "Sample"
+    Workflow.require_workflow "HTS"
+  end
 
-  def _test_template
+  def __test_template
     job = Sample.job(:mutect2, "small", :reference => "hg38")
 
     TmpFile.with_file do |batch_dir|
@@ -17,13 +19,13 @@ class TestSLURM < Test::Unit::TestCase
     end
   end
 
-  def test_run_job
+  def __test_run_job
     job = Sample.job(:mutect2, "small", :reference => "hg38")
 
     job.clean
 
     jobid = HPC::SLURM.run_job(job, :workflows => "HTS", :batch_modules => 'java', :env_cmd => '_JAVA_OPTIONS="-Xms1g -Xmx${MAX_MEMORY}m"', :queue => :debug, :time => '01:00:00', :config_keys => "HTS_light", :task_cpus => '10', :tail => true, :clean_task => "HTS#mutect2")
-    iii jobid
+    assert jobid.to_s =~ /^\d+$/
   end
 
 end

data/test/rbbt/knowledge_base/test_enrichment.rb

@@ -6,35 +6,41 @@ require 'rbbt/knowledge_base/enrichment'
 
 
 class TestKnowledgeBaseEnrichment < Test::Unit::TestCase
+  def knowledge_base(file)
 
-  EFFECT =StringIO.new <<-END
+    effect =StringIO.new <<-END
 #: :sep=" "#:type=:double
 #SG TG Effect
 MDM2 TP53 inhibition
 TP53 NFKB1|GLI1 activation|activation true|true
-  END
-
-  EFFECT_OPTIONS = {
-    :source => "SG=~Associated Gene Name",
-    :target => "TG=~Associated Gene Name=>Ensembl Gene ID",
-    :persist => false,
-    :identifiers => datafile_test('identifiers'),
-    :undirected => true,
-    :namespace => "Hsa"
-  }
-
-  EFFECT_TSV = TSV.open EFFECT, EFFECT_OPTIONS.dup 
-
-  KNOWLEDGE_BASE = KnowledgeBase.new '/tmp/kb.foo', "Hsa/feb2014"
-  KNOWLEDGE_BASE.format = {"Gene" => "Ensembl Gene ID"}
-
-  KNOWLEDGE_BASE.register :effects, EFFECT_TSV, EFFECT_OPTIONS.dup
-  KNOWLEDGE_BASE.register :gene_ages, datafile_test('gene_ages')
-  KNOWLEDGE_BASE.register :nature, datafile_test('nature'), :source => "UniProt/SwissProt Accession", :target => "NCI Nature Pathway ID"
-  
+    END
+
+    effect_options = {
+      :source => "SG=~Associated Gene Name",
+      :target => "TG=~Associated Gene Name=>Ensembl Gene ID",
+      :persist => false,
+      :identifiers => datafile_test('identifiers'),
+      :undirected => true,
+      :namespace => "Hsa"
+    }
+
+    effect_tsv = TSV.open effect, effect_options.dup 
+
+    knowledge_base = KnowledgeBase.new file, "Hsa/feb2014"
+    knowledge_base.format = {"Gene" => "Ensembl Gene ID"}
+
+    knowledge_base.register :effects, effect_tsv, effect_options.dup
+    knowledge_base.register :gene_ages, datafile_test('gene_ages')
+    knowledge_base.register :nature, datafile_test('nature'), :source => "UniProt/SwissProt Accession", :target => "NCI Nature Pathway ID"
+
+    knowledge_base
+  end
+
   def test_enrichment
-    genes = %w(P17706-2 LMAN1 P17706-1 P29353-2 JAK3 Q8NFM1 EIF2AK2 JAK1 SRC PIAS1 KPNB1 KPNA2 STAT3 PTPN1 ATR CREBBP PTPRA SGK1 P46108-1 NCK2 O00145 PTPN1 CBL SORBS1 IRS1 SHC1 AKT2 GRB2 SOS1 RPS6KB1 AKT1 DOK1 RASA1 NCK1 FOXO3 RAPGEF1 TRIP10 EIF4EBP1 PDPK1 GRB14 PTPN11 INS GRB10 CAV1 EIF4B RPS6KB1 EEF2 EEF2K AKT1 TSC1 TSC2 CLIP1 DDIT4 SGK1 PDPK1 DEPTOR SREBF1 CYCS IRS1 RPS6KA1 BNIP3 RRN3 RICTOR IKBKB AKT1S1 PXN PML EIF4A1 PPARGC1A YY1 PRKCA RPTOR PDCD4 SIK1 P10636-8 BRSK1 MYC SMARCD3 STK11 ETV4 MARK4 MAP2 MARK2 CRTC2 PSEN2 MST4 CTSD BRSK2 SIK2 ESR1 CAB39 STK11IP SMAD4 CREB1 PRKACA EZR TP53 GSK3B SIK3 CDC37 HSP90AA1 )
-    assert KNOWLEDGE_BASE.enrichment(:nature, genes).any?
+    TmpFile.with_file do |tmp_kb|
+      genes = %w(P17706-2 LMAN1 P17706-1 P29353-2 JAK3 Q8NFM1 EIF2AK2 JAK1 SRC PIAS1 KPNB1 KPNA2 STAT3 PTPN1 ATR CREBBP PTPRA SGK1 P46108-1 NCK2 O00145 PTPN1 CBL SORBS1 IRS1 SHC1 AKT2 GRB2 SOS1 RPS6KB1 AKT1 DOK1 RASA1 NCK1 FOXO3 RAPGEF1 TRIP10 EIF4EBP1 PDPK1 GRB14 PTPN11 INS GRB10 CAV1 EIF4B RPS6KB1 EEF2 EEF2K AKT1 TSC1 TSC2 CLIP1 DDIT4 SGK1 PDPK1 DEPTOR SREBF1 CYCS IRS1 RPS6KA1 BNIP3 RRN3 RICTOR IKBKB AKT1S1 PXN PML EIF4A1 PPARGC1A YY1 PRKCA RPTOR PDCD4 SIK1 P10636-8 BRSK1 MYC SMARCD3 STK11 ETV4 MARK4 MAP2 MARK2 CRTC2 PSEN2 MST4 CTSD BRSK2 SIK2 ESR1 CAB39 STK11IP SMAD4 CREB1 PRKACA EZR TP53 GSK3B SIK3 CDC37 HSP90AA1 )
+      assert knowledge_base(tmp_kb).enrichment(:nature, genes).any?
+    end
   end
 
 end

data/test/rbbt/knowledge_base/test_traverse.rb

@@ -2,8 +2,11 @@ require File.join(File.expand_path(File.dirname(__FILE__)), '../..', 'test_helpe
 require 'rbbt/knowledge_base/traverse'
 require 'rbbt/workflow'
 
-Workflow.require_workflow "Genomics"
 class TestKnowledgeBaseTraverse < Test::Unit::TestCase
+  def setup
+    Workflow.require_workflow "Genomics"
+  end
+
   def kb
     Genomics.knowledge_base
   end

data/test/rbbt/test_knowledge_base.rb

@@ -9,15 +9,8 @@ require 'rbbt/entity/identifiers'
 require 'rbbt/association'
 require 'rbbt/knowledge_base'
 
-require 'rbbt/sources/organism'
-require 'rbbt/sources/tfacts'
-require 'rbbt/sources/kegg'
-
 module Gene
   extend Entity
-  add_identifiers Organism.identifiers("NAMESPACE"), "Ensembl Gene ID", "Associated Gene Name"
-  add_identifiers KEGG.identifiers
-
   property :follow => :single do |kb,name,annotate=nil|
     if annotate.nil? or annotate
       l = kb.children(name, self).target_entity
@@ -55,6 +48,15 @@ end
 
 
 class TestKnowledgeBase < Test::Unit::TestCase
+  def setup
+    require 'rbbt/sources/organism'
+    require 'rbbt/sources/tfacts'
+    require 'rbbt/sources/kegg'
+
+    Gene.add_identifiers Organism.identifiers("NAMESPACE"), "Ensembl Gene ID", "Associated Gene Name"
+    Gene.add_identifiers KEGG.identifiers
+
+  end
 
   def test_knowledge_base_reverse
     organism = Organism.default_code("Hsa")

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-util
 version: !ruby/object:Gem::Version
-  version: 5.37.8
+  version: 5.37.9
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2023-03-14 00:00:00.000000000 Z
+date: 2023-03-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake
@@ -585,7 +585,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.1.4
+rubygems_version: 3.4.8
 signing_key: 
 specification_version: 4
 summary: Utilities for the Ruby Bioinformatics Toolkit (rbbt)