rbbt-sources 3.1.46 → 3.1.51

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 9cfdfc8f6fd6a11106d90cf39dc8f10165fd36128fe69a8ca4c72fcf878183c3
-  data.tar.gz: f268c37d7c50b74fb67d2dae77bf25c93dbf1013f23a11e1ffeb0f024fdabdc0
+  metadata.gz: 4cb92f1a2c300c6787870f06f5b4ea45967242a33e7c51cf94f002d1901000af
+  data.tar.gz: 793c56312f02532861451142988b24463de6a30d460f6f7a7238889b91c9c336
 SHA512:
-  metadata.gz: d6e29cb4355ad3307bf3ccbead098406e51216e963cd201fc11c17aa4aad263d184ca1ab1389c4940e32bf1d541471a9aac326b12055585b515f26c63592a025
-  data.tar.gz: 4663061a1b3f8a80914468c8af3bb8b5da821739fe8943011881ba9b89405e34b40ec9b145e9a7f9fdf7b2cc7cdf8f294fa978b01c41b1f8693a4e73b186d65f
+  metadata.gz: 95ab052b23bc28f919e9cf242e0a74e1e433335879d77fc38752925f74e5ec92b8b42b1c27f1fee170bb3125be1a04b83b24c7da524c46efce07241ff169bdff
+  data.tar.gz: bbe1970028942398b1b84e314d704acb0fac405045785e45516e8dbd65927852874bcceb0393748fd68a507c6d4292fda579d1df16d7831d47b5b9ea0de6bf7d

data/etc/biomart/missing_in_archive

@@ -1,3 +1,7 @@
+">oct2014":
+    - rgd~rgd_id
+">jun2019":
+    - entrezgene~entrezgene_id
 ">dec2017":
     - unigene
 ">dec2016":

data/lib/rbbt/sources/biomart.rb

@@ -134,11 +134,14 @@ module BioMart
   # cause an error if the BioMart WS does not allow filtering with that
   # attribute.
   def self.query(database, main, attrs = nil, filters = nil, data = nil, open_options = {})
+    IndiferentHash.setup(open_options)
     open_options = Misc.add_defaults open_options, :nocache => false, :filename => nil, :field_names => nil, :by_chr => false
     filename, field_names, by_chr = Misc.process_options open_options, :filename, :field_names, :by_chr
     attrs   ||= []
     open_options = Misc.add_defaults open_options, :keep_empty => false, :merge => true
 
+    IndiferentHash.setup(open_options)
+
     Log.low "BioMart query: '#{main}' [#{(attrs || []) * ', '}] [#{(filters || []) * ', '}] #{open_options.inspect}"
 
     max_items = 2
@@ -167,6 +170,7 @@ module BioMart
     end
 
     open_options[:filename] = "BioMart[#{main}+#{attrs.length}]"
+
     if filename.nil?
       results = TSV.open data, open_options
       results.key_field = main

data/lib/rbbt/sources/ensembl_ftp.rb

@@ -52,24 +52,35 @@ module Ensembl
       File.join("ftp://" + SERVER, ftp_directory_for(organism) )
     end
 
-    def self.url_for(organism, table)
-      "#{base_url(organism)}/#{table}.txt.gz.bz2"
+    def self.url_for(organism, table, extension)
+      File.join(base_url(organism), table) + ".#{extension}.gz"
+    end
+
+    def self._get_gz(url)
+      begin
+        CMD.cmd("wget '#{url}' -O  - | gunzip").read
+      rescue
+        CMD.cmd("wget '#{url}.bz2' -O  - | bunzip2 | gunzip").read
+      end
+    end
+
+    def self._get_file(organism, table, extension)
+      url = url_for(organism, table, extension)
+      self._get_gz(url)
     end
 
     def self.has_table?(organism, table)
-      sql_file = CMD.cmd("wget '#{base_url(organism)}/#{File.basename(base_url(organism))}.sql.gz.bz2' -O  -| bunzip2| gunzip").read
+      sql_file = _get_file(organism, File.basename(base_url(organism)), 'sql')
       ! sql_file.match(/^CREATE TABLE .#{table}. \((.*?)^\)/sm).nil?
     end
 
     def self.fields_for(organism, table)
-      sql_file = CMD.cmd("wget '#{base_url(organism)}/#{File.basename(base_url(organism))}.sql.gz.bz2' -O  -| bunzip2| gunzip").read
-      
+      sql_file = _get_file(organism, File.basename(base_url(organism)), 'sql')
       chunk = sql_file.match(/^CREATE TABLE .#{table}. \((.*?)^\)/sm)[1]
       chunk.scan(/^\s+`(.*?)`/).flatten
     end
 
     def self.ensembl_tsv(organism, table, key_field = nil, fields = nil, options = {})
-      url = url_for(organism, table)
       if key_field and fields
         all_fields = fields_for(organism, table)
         key_pos = all_fields.index key_field
@@ -78,7 +89,8 @@ module Ensembl
         options[:key_field] = key_pos
         options[:fields]    = field_pos
       end
-      tsv = TSV.open(CMD.cmd("wget '#{url}' -O - |bunzip2|gunzip", :pipe => true), options)
+
+      tsv = TSV.open(StringIO.new(_get_file(organism, table, "txt")), options)
       tsv.key_field = key_field
       tsv.fields = fields
       tsv

data/lib/rbbt/sources/organism.rb

@@ -88,8 +88,10 @@ module Organism
       end
     when "Mmu"
       "mm10"
+    when "Rno"
+      "rn6"
     else
-      raise "Only organism 'Hsa' (Homo sapiens) and Mmu (Mus musculus) supported" 
+      raise "Only organism 'Hsa' (Homo sapiens), 'Rno' (Rattus norvegicus), and Mmu (Mus musculus) supported" 
     end
   end
 
@@ -117,6 +119,8 @@ module Organism
       end
     when "Mmu"
       "GRCm38"
+    when "Rno"
+      "Rnor_6.0"
     else
       raise "Only organism 'Hsa' (Homo sapiens) and Mmu (Mus musculus) supported" 
     end
@@ -145,12 +149,12 @@ module Organism
     new_positions = {}
 
     TmpFile.with_file(positions_bed) do |source_bed|
-      TmpFile.with_file() do |unmapped_file|
-        TmpFile.with_file() do |map_file|
+      TmpFile.with_file do |unmapped_file|
+        TmpFile.with_file do |map_file|
 
 
           Open.write(map_file, Open.read(map_url))
-          new_mutations = TmpFile.with_file() do |target_bed|
+          new_mutations = TmpFile.with_file do |target_bed|
             FileUtils.chmod(755, Rbbt.software.opt.bin.liftOver.produce.find)
             CMD.cmd("#{Rbbt.software.opt.bin.liftOver.find} '#{source_bed}' '#{map_file}' '#{target_bed}' '#{unmapped_file}'").read
             Open.read(target_bed) do |line|

data/lib/rbbt/sources/uniprot.rb

@@ -203,12 +203,19 @@ module UniProt
       end
       value = part.gsub("\nFT", '').gsub(/\s+/, ' ')
       case
+      when value.match(/(\d+)\.\.(\d+) (.*)/)
+        start, eend, description = $1, $2, $3
+        description.gsub(/^FT\s+/m, '')
       when value.match(/(\d+) (\d+) (.*)/)
         start, eend, description = $1, $2, $3
         description.gsub(/^FT\s+/m, '')
       when value.match(/^\s+(\d+) (\d+)/)
         start, eend = $1, $2
         description = nil
+      when value.match(/(\d+) (.*)/)
+        start, description = $1, $2
+        eend = start
+        description.gsub(/^FT\s+/m, '')
       else
         Log.debug "Value not understood: #{ value }"
       end

data/share/Ensembl/release_dates

@@ -1,5 +1,10 @@
 #: :type=:single
 #Release	build
+release-100	apr2020
+release-99	jan2020
+release-98	sep2019
+release-97	jul2019
+release-96	apr2019
 release-95	jan2019
 release-94	oct2018
 release-93	jul2018

data/share/install/Organism/Hsa/Rakefile

@@ -5,7 +5,7 @@ require File.join(File.dirname(__FILE__), '../../lib/helpers')
 
 $taxs = [9606]
 $scientific_name = "Homo sapiens"
-$ortholog_key = "human_ensembl_gene"
+$ortholog_key = "hsapiens_homolog_ensembl_gene"
 
 $biomart_db = 'hsapiens_gene_ensembl'
 $biomart_db_germline_variation = 'hsapiens_snp'

data/share/install/Organism/Mmu/Rakefile

@@ -5,7 +5,7 @@ require File.join(File.dirname(__FILE__), '../../lib/helpers')
 
 $taxs = [10090]
 $scientific_name = "Mus musculus"
-$ortholog_key = "mouse_ensembl_gene"
+$ortholog_key = "mmusculus_homolog_ensembl_gene"
 
 $biomart_db = 'mmusculus_gene_ensembl'
 $biomart_db_germline_variation = 'mmusculus_snp'

data/share/install/Organism/Rno/Rakefile

@@ -9,7 +9,7 @@ $scientific_name = "Rattus norvegicus"
 $biomart_db = 'rnorvegicus_gene_ensembl'
 $biomart_db_germline_variation = 'rnorvegicus_snp'
 $biomart_db_somatic_variation = 'rnorvegicus_snp_som'
-$ortholog_key = "rat_ensembl_gene"
+$ortholog_key = "rnorvegicus_homolog_ensembl_gene"
 
 $biomart_lexicon = [ 
   [ 'Associated Gene Name' , "external_gene_id"], 
@@ -20,6 +20,7 @@ $biomart_lexicon = [
 
 $biomart_identifiers = [ 
   ['Entrez Gene ID', "entrezgene"],
+  ['Ensembl Protein ID', "ensembl_peptide_id"  ],
   ['Associated Gene Name' , "rgd_symbol"], 
   ['Protein ID' , "protein_id"] , 
   ['UniProt/SwissProt ID' , "uniprot_swissprot"] , 

data/share/install/Organism/organism_helpers.rb

@@ -547,13 +547,13 @@ end
 rule /^possible_ortholog_(.*)/ do |t|
   other = t.name.match(/ortholog_(.*)/)[1]
   other_key = Organism.ortholog_key(other).produce.read
-  BioMart.tsv($biomart_db, $biomart_ensembl_gene, [["Ortholog Ensembl Gene ID", "inter_paralog_" + other_key]], [], nil, :keep_empty => false, :type => :flat, :filename => t.name, :namespace => Thread.current['namespace'])
+  BioMart.tsv($biomart_db, $biomart_ensembl_gene, [["Ortholog Ensembl Gene ID", "inter_paralog_" + other_key]], [], nil, :keep_empty => false, :type => :double, :filename => t.name, :namespace => Thread.current['namespace'])
 end
 
 rule /^ortholog_(.*)/ do |t|
   other = t.name.match(/ortholog_(.*)/)[1]
   other_key = Organism.ortholog_key(other).produce.read
-  BioMart.tsv($biomart_db, $biomart_ensembl_gene, [["Ortholog Ensembl Gene ID", other_key]], [], nil, :keep_empty => false, :type => :flat, :filename => t.name, :namespace => Thread.current['namespace'])
+  BioMart.tsv($biomart_db, $biomart_ensembl_gene, [["Ortholog Ensembl Gene ID", other_key]], [], nil, :keep_empty => false, :type => :double, :filename => t.name, :namespace => Thread.current['namespace'])
 end
 
 rule /[a-z]{3}[0-9]{4}\/.*/i do |t|
@@ -728,13 +728,18 @@ file 'protein_sequence' => ["transcripts", "transcript_5utr", "transcript_3utr",
   transcript_sequence.through do |transcript, sequence|
     protein = transcript_protein[transcript]
     next if protein.nil? or protein.empty?
+
     utr5 = transcript_5utr[transcript]
     utr3 = transcript_3utr[transcript]
     phase = transcript_phase[transcript] || 0
+
     if phase < 0
-      utr5 = - phase if utr5 == 0
+      if utr5.nil? || utr5 == 0 || utr5 == "0"
+        utr5 = 0
+      end
       phase = 0
     end
+
     psequence = Bio::Sequence::NA.new(("N" * phase) << sequence[utr5..sequence.length-utr3-1]).translate
     protein_sequence[protein]=psequence
   end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-sources
 version: !ruby/object:Gem::Version
-  version: 3.1.46
+  version: 3.1.51
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-05-14 00:00:00.000000000 Z
+date: 2021-01-28 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rbbt-util