rbbt-util 5.33.15 → 5.33.18

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c0573e1c183e61a16493ee1c9237fdc20901d33eca6713be8558d01a0014068e
-  data.tar.gz: b365435a324ed8fb7a3583ccd9904a968a93d5e655e96f09248cf1b7f6216780
+  metadata.gz: 972b137af2c124d5288389d49e188d3973246a912c0405c391cdb6b8a5987cae
+  data.tar.gz: 4744dc8f03a8230ad8d5f61f1d15f14c4e5635d5e3798be584083217269b88a5
 SHA512:
-  metadata.gz: 4dedeeffa67b1a95197188a7db86ee25f8036c04c5caa36d56c421fe739f85449af010ab342e0711d8d204a675e628bb9ba646ac3317e2d15a73103600d6d63e
-  data.tar.gz: a3f8812b2b46a98dc0ce150667f8b08ceea415f1e7e765d25164cd9af68a0dffef9e33bdb5e73e3408e137195d7b827465484351cf51e89ef0cb01237134b887
+  metadata.gz: 34f7cbdd329d23341cc6591bfe19f584a4c50862c8fae6c2ccd8d72bcd76fbb129acbaa3bbb03e08eb633aa475d5d0eacb1a015e667db169ebf1f09e588ab579
+  data.tar.gz: 8110363114c606214d115ffb6475ce277dc3f605c1e8eafd97911a53cc259386ed6173d45cca4a5692b8e4ca0a89db71a77b6fb5eca69400b565b820a83a8d56

data/lib/rbbt/tsv/util.rb

@@ -1,12 +1,39 @@
 require 'rbbt/resource/path'
 module TSV
 
+  def self.stream_column(file, column)
+    header = TSV.parse_header(file)
+    pos = header.fields.index(column) + 1
+    sep2 = header.options[:sep2] || "|"
+    case header.type.to_s
+    when nil, "double"
+      TSV.traverse file, :type => :array, :into => :stream do |line|
+        next if line =~ /^#/
+        line.split("\t")[pos].gsub(sep2, "\n")
+      end
+    when "single"
+      TSV.traverse file, :type => :array, :into => :stream do |line|
+        next if line =~ /^#/
+        line.split("\t")[1]
+      end
+    when "flat"
+      TSV.traverse file, :type => :array, :into => :stream do |line|
+        next if line =~ /^#/
+        line.split("\t")[1..-1] * "\n"
+      end
+    when 'list'
+      TSV.traverse file, :type => :array, :into => :stream do |line|
+        next if line =~ /^#/
+        line.split("\t")[pos]
+      end
+    end
+  end
+
   def self.guess_id(identifier_file, values, options = {})
     field_matches = TSV.field_match_counts(identifier_file, values, options)
     field_matches.sort_by{|field, count| count.to_i}.last
   end
 
-
   def self.field_match_counts(file, values, options = {})
     options = Misc.add_defaults options, :persist_prefix => "Field_Matches"
     persist_options = Misc.pull_keys options, :persist

data/lib/rbbt/util/misc/omics.rb

@@ -203,33 +203,48 @@ module Misc
 
 
   def self.translate_dna_mutation_hgvs2rbbt(cds)
-    change = case
-             when cds =~ />/
-               cds.split(">").last
-             when cds =~ /del/
-               deletion = cds.split("del").last.chomp
+    change = begin
                case
-               when deletion =~ /^\d+$/
-                 "-" * deletion.to_i
-               when deletion =~ /^[ACTG]+$/i
-                 "-" * deletion.length
-               else
-                 Log.debug "Unknown deletion: #{ deletion }"
-                 deletion
-               end
-             when cds =~ /ins/
-               insertion = cds.split("ins").last
-               case
-               when insertion =~ /^\d+$/
-                 "+" + "N" * insertion.to_i
-               when insertion =~ /^[NACTG]+$/i
-                 "+" + insertion
+               when cds =~ />/
+                 cds.split(">").last
+               when cds =~ /delins/
+                 del, ins = cds.split("delins")
+                 start, eend = del.split("_")
+                 del_size = eend.to_i - start.to_i + 1
+                 if ins =~ /^[ACTG]+$/i
+                   ("-" * del_size) + ins
+                 else
+                   Log.debug "Unknown delins: #{ cds }"
+                   "?(" << cds << ")"
+                 end
+               when cds =~ /del/
+                 deletion = cds.split("del").last.chomp
+                 case
+                 when deletion =~ /^\d+$/
+                   "-" * deletion.to_i
+                 when deletion =~ /^[ACTG]+$/i
+                   "-" * deletion.length
+                 else
+                   Log.debug "Unknown deletion: #{ cds }"
+                   "?(" << cds << ")"
+                 end
+               when cds =~ /ins/
+                 insertion = cds.split("ins").last
+                 case
+                 when insertion =~ /^\d+$/
+                   "+" + "N" * insertion.to_i
+                 when insertion =~ /^[NACTG]+$/i
+                   "+" + insertion
+                 else
+                   Log.debug "Unknown insertion: #{cds }"
+                   "?(" << cds << ")"
+                 end
                else
-                 Log.debug "Unknown insertion: #{insertion }"
-                 insertion
+                 Log.debug "Unknown change: #{cds}"
+                 "?(" << cds << ")"
                end
-             else
-               Log.debug "Unknown change: #{cds}"
+             rescue
+               Log.debug "Error processing change: #{cds}"
                "?(" << cds << ")"
              end
     change
@@ -323,11 +338,11 @@ module Misc
       line.partition(sep).last
     end
   end
-  
+
   def self.sort_genomic_locations_strict(stream, sep = ":")
     sort_stream(stream, '#', "-k1,1V -k2,2n -t#{sep}")
   end
-  
+
   def self.sort_genomic_locations(stream, sep = ":")
     sort_stream(stream, '#', "-k1,1 -k2,2n -t#{sep}")
   end
@@ -483,7 +498,7 @@ module Misc
           sin << [chr, start, eend, id] * "\t" << "\n"
         end
       end
-      
+
       TmpFile.with_file do |tmpfile|
         Misc.consume_stream(nio, false, tmpfile)
 

data/lib/rbbt/util/misc/system.rb

@@ -94,7 +94,7 @@ module Misc
   end
 
   def self.is_filename?(string, need_to_exists = true)
-    return false if string.nil?
+    return false if string.nil? or string.include? "\n"
     return true if defined? Path and Path === string
     return true if string.respond_to? :exists
     return true if String === string and string.split("/").select{|p| p.length > 265}.empty? and (! need_to_exists || File.exist?(string))

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

@@ -145,7 +145,11 @@ class Step
   end
 
   def input_dependencies
-    @input_dependencies ||= (recursive_inputs.flatten.select{|i| Step === i } + recursive_inputs.flatten.select{|dep| Path === dep && Step === dep.resource }.collect{|dep| dep.resource })
+    @input_dependencies ||= (recursive_inputs.flatten.select{|i| Step === i } + 
+                             recursive_inputs.flatten.
+                              select{|dep| Path === dep && Step === dep.resource }.
+                              select{|dep| ! dep.resource.started? }. # Ignore input_deps already started
+                              collect{|dep| dep.resource })
   end
 
   def execute_dependency(dependency, log = true)

data/lib/rbbt/workflow/usage.rb

@@ -125,6 +125,8 @@ module Workflow
       first = last.nil?
       last = _prov_tasks(workflow.dep_tree(task_name))
 
+      break if child
+
       if child
         description << "-> " << task_name.to_s
       elsif first

data/test/rbbt/util/misc/test_omics.rb

@@ -16,22 +16,22 @@ class TestMiscOmics < Test::Unit::TestCase
     assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.209+1delGTAA"), "----"
     assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.371_397del27"), "---------------------------"
     assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.1359+2_1359+11delTTAGAAGAGC"), "----------"
-    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.ins?"), "?"
-    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.?_?ins?"), "?"
-    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.184_185ins?"), "?"
+    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.ins?"), "?(c.ins?)"
+    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.?_?ins?"), "?(c.?_?ins?)"
+    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.184_185ins?"), "?(c.184_185ins?)"
     assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.?_?ins57"), "+NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN"
     assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.?_?insCT"), "+CT"
-    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.?_?del?"), "?"
+    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.?_?del?"), "?(c.?_?del?)"
     assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.?"), "?(c.?)"
     assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.(312)?"), "?(c.(312)?)"
     assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.(1669_1671)>?"), "?"
     assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.1050-?"), "?(c.1050-?)"
     assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.(1347_1540)?"), "?(c.(1347_1540)?)"
-    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.1-?_850+?del"), "c.1-?_850+?"
-    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.2289_?del(7236)"), "(7236)"
-    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.(3916_3927)del?"), "?"
-    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.1705_?del?"), "?"
-    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.1-?_421+?del"), "c.1-?_421+?"
+    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.1-?_850+?del"), "?(c.1-?_850+?del)"
+    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.2289_?del(7236)"), "?(c.2289_?del(7236))"
+    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.(3916_3927)del?"), "?(c.(3916_3927)del?)"
+    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.1705_?del?"), "?(c.1705_?del?)"
+    assert_equal Misc.translate_dna_mutation_hgvs2rbbt("c.1-?_421+?del"), "?(c.1-?_421+?del)"
   end
 
   def test_translate_prot_mutation_hgvs2rbbt

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-util
 version: !ruby/object:Gem::Version
-  version: 5.33.15
+  version: 5.33.18
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2022-05-05 00:00:00.000000000 Z
+date: 2022-05-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake