rbbt-sources 3.1.29 → 3.1.30

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 14ec519ad1a7309351fadacf8d418b197c931769
-  data.tar.gz: 8426ce938123e929a04a07bbb6c0ab5d12be6b04
+  metadata.gz: 7da6b2938f54788ddcd4ee03b78e8e8b237bf478
+  data.tar.gz: fc847e1ddf84cb477b4f07ac8b33a393009e72ac
 SHA512:
-  metadata.gz: d817ada6d2df2a355dd3ac7ae6c98d3261f029bd9a82a02e45c0022118ff88b99e5ead63c6d424c6a522e0c4d89f66ac4b0c00c33d4007cd9462dbb8b3923ba6
-  data.tar.gz: adf03a5ba5263bcb10877e8823826b9239cbabc73e9fe81ace8b2d21e2da0505d6a350f7c36f44efe73af0257863ec0077b6a7255abbb7daa3dfba857b6717bc
+  metadata.gz: d2a2261e40edd858ca483b269d409a95e48f61acfeb07c9b31e2381e5bcffa619c06c6c5afaa4df05ba91405638bbaa94897722c66ef1610344b542746c2aa61
+  data.tar.gz: 7bd00cb785efd5db78ba14607436e42465055da71119d27741f20b7c7de4b791b90c321aaecc1de1d681c9d5ba75823401d03fd5da247d7a187dec56a27c4fd3

data/lib/rbbt/sources/COREAD_phospho_proteome.rb

@@ -0,0 +1,183 @@
+require 'rbbt-util'
+require 'rbbt/resource'
+
+# CITE: Genomic Determinants of Protein Abundance Variation in Colorectal
+# Cancer Cells PMID: 28854368
+#
+# Roumeliotis TI, Williams SP, Gonçalves E, et al. Genomic Determinants of
+# Protein Abundance Variation in Colorectal Cancer Cells. Cell Reports.
+# 2017;20(9):2201-2214. doi:10.1016/j.celrep.2017.08.010.
+
+module COREADPhosphoProteome
+  extend Resource
+  self.subdir = 'share/databases/COREADPhosphoProteome'
+
+  #def self.organism(org="Hsa")
+  #  Organism.default_code(org)
+  #end
+
+  #self.search_paths = {}
+  #self.search_paths[:default] = :lib
+
+  
+
+  COREADPhosphoProteome.claim COREADPhosphoProteome[".source/mmc3.xlsx"], :url, "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5583477/bin/mmc3.xlsx"
+
+  COREADPhosphoProteome.claim COREADPhosphoProteome.data, :proc do 
+    require 'rbbt/tsv/excel'
+    io = TSV.excel COREADPhosphoProteome[".source/mmc3.xlsx"].produce.find, :text => true
+    TSV.collapse_stream io
+  end
+
+  COREADPhosphoProteome.claim COREADPhosphoProteome.phosphosite_levels, :proc do 
+    tsv = COREADPhosphoProteome.data.tsv 
+    name, seq, site, kinases, name, *cell_lines = tsv.fields 
+    tsv.add_field "Phosphosite" do |uni,values_list|
+      Misc.zip_fields(values_list).collect{|values|
+        name, seq, site, kinases, kegg_name, *vals = values 
+        [name, site] * ":"
+      }
+    end
+    tsv.reorder "Phosphosite", cell_lines, :zipped => true
+  end
+
+  COREADPhosphoProteome.claim COREADPhosphoProteome.phosphosite_binary, :proc do
+    require 'rbbt/matrix'
+    require 'rbbt/matrix/barcode'
+
+    m = RbbtMatrix.new COREADPhosphoProteome.phosphosite_levels.find
+    a = m.to_activity(3).tsv(false)
+    a
+  end
+
+  COREADPhosphoProteome.claim COREADPhosphoProteome.signor_activity_present, :proc do
+    require 'rbbt/sources/signor'
+    signor = Signor.phospho_sites.tsv
+
+    signor.add_field "Fixed" do |k,l|
+      case l.uniq.length 
+      when 1
+        l.first
+      when 2
+        "Unclear"
+      end
+    end
+    signor = signor.slice("Fixed").to_single
+
+    parser = TSV::Parser.new COREADPhosphoProteome.phosphosite_levels
+    dumper = TSV::Dumper.new parser.options
+    dumper.init
+    cell_lines = parser.fields
+    TSV.traverse parser, :into => dumper do |site,values|
+      osite = site
+      site = site.sub(':S', ':Ser').sub(':T', ':Thr').sub(':Y', ':Tyr')
+      next unless signor.include? site
+      new_values = values.flatten.zip(cell_lines).collect{|value,cell_line|
+        next if signor[site] == "Unclear"
+        case value
+        when nil, ""
+          signor[site] == "Activates" ? -1 : 1
+        else
+          signor[site] == "Activates" ? 1 : -1
+        end
+      }
+      [site, new_values]
+    end
+  end
+
+  COREADPhosphoProteome.claim COREADPhosphoProteome.signor_activity_100, :proc do
+    require 'rbbt/sources/signor'
+    signor = Signor.phospho_sites.tsv
+
+    signor.add_field "Fixed" do |k,l|
+      case l.uniq.length 
+      when 1
+        l.first
+      when 2
+        "Unclear"
+      end
+    end
+    signor = signor.slice("Fixed").to_single
+
+    parser = TSV::Parser.new COREADPhosphoProteome.phosphosite_levels
+    dumper = TSV::Dumper.new parser.options
+    dumper.init
+    TSV.traverse parser, :into => dumper do |site,values|
+      osite = site
+      site = site.sub(':S', ':Ser').sub(':T', ':Thr').sub(':Y', ':Tyr')
+      next unless signor.include? site
+      new_values = values.flatten.collect{|value|
+        next if signor[site] == "Unclear"
+        case value
+        when nil, ""
+          signor[site] == "Activates" ? -1 : 1
+        else
+          if value.to_f >= 100
+            signor[site] == "Activates" ? 1 : -1
+          else
+            signor[site] == "Activates" ? -1 : 1
+          end
+        end
+      }
+      [site, new_values]
+    end
+  end
+
+  COREADPhosphoProteome.claim COREADPhosphoProteome.signor_activity_levels, :proc do
+    require 'rbbt/sources/signor'
+    signor = Signor.phospho_sites.tsv
+
+    signor.add_field "Fixed" do |k,l|
+      case l.uniq.length 
+      when 1
+        l.first
+      when 2
+        "Unclear"
+      end
+    end
+    signor = signor.slice("Fixed").to_single
+
+
+    parser = TSV::Parser.new COREADPhosphoProteome.phosphosite_binary
+    dumper = TSV::Dumper.new parser.options
+    dumper.init
+    TSV.traverse parser, :into => dumper do |site,values|
+      osite = site
+      site = site.first if Array === site
+      site = site.sub(':S', ':Ser').sub(':T', ':Thr').sub(':Y', ':Tyr')
+      next unless signor.include? site
+      max = values.flatten.max
+      new_values = values.flatten.collect{|value|
+        next if signor[site] == "Unclear"
+        case value
+        when nil, ""
+          signor[site] == "Activates" ? -1 : 1
+        else
+          if value == max
+            signor[site] == "Activates" ? 1 : -1
+          else
+            signor[site] == "Activates" ? -1 : 1
+          end
+        end
+      }
+      [site, new_values]
+    end
+  end
+
+  COREADPhosphoProteome.claim COREADPhosphoProteome.cascade_levels, :proc do
+    require 'rbbt/sources/CASCADE'
+
+    cascade_proteins = CASCADE.members.tsv.values.flatten.compact.uniq
+    tsv = COREADPhosphoProteome.phosphosite_levels.tsv
+    tsv.select do |site,values|
+      cascade_proteins.include? site.split(":").first
+    end
+  end
+end
+
+iif COREADPhosphoProteome.data.produce.find if __FILE__ == $0
+iif COREADPhosphoProteome.phosphosite_levels.produce.find if __FILE__ == $0
+iif COREADPhosphoProteome.phosphosite_binary.produce.find if __FILE__ == $0
+iif COREADPhosphoProteome.signor_activity_present.produce(true).find if __FILE__ == $0
+iif COREADPhosphoProteome.cascade_levels.produce.find if __FILE__ == $0
+

data/lib/rbbt/sources/array_express.rb

@@ -0,0 +1,17 @@
+module ArrayExpress
+
+  def self.tpm(code, organism = Organism.default_code("Hsa"))
+    url = "https://www.ebi.ac.uk/gxa/experiments-content/#{code}/resources/ExperimentDownloadSupplier.RnaSeqBaseline/tpms.tsv"
+    io = TSV.traverse Open.open(url), :type => :line, :into => :stream do |line|
+      next if line =~ /^#/
+      parts = line.split("\t")
+      line = parts[0] << "\t" << parts[2..-1] * "\t"
+      line = "#" + line if line =~ /Gene ID/
+      line
+    end
+    tsv = TSV.open(io, :type => :list, :cast => :to_f)
+    tsv.key_field = "Ensembl Gene ID"
+    tsv.namespace = organism
+    tsv
+  end
+end

data/lib/rbbt/sources/phospho_site_plus.rb

@@ -1,9 +1,9 @@
 require 'rbbt-util'
 require 'rbbt/resource'
 
-module PhosphoSitePlues
+module PhosphoSitePlus
   extend Resource
-  self.subdir = 'share/databases/PhosphoSitePlues'
+  self.subdir = 'share/databases/PhosphoSitePlus'
 
   def self.organism(org="Hsa")
     require 'rbbt/sources/organism'
@@ -18,7 +18,7 @@ module PhosphoSitePlues
 
 
 
-  ALL_FILES = %(Acetylation_site_dataset.gz Disease-associated_sites.gz
+  ALL_FILES = %w(Acetylation_site_dataset.gz Disease-associated_sites.gz
 Kinase_Substrate_Dataset.gz Methylation_site_dataset.gz
 O-GalNAc_site_dataset.gz O-GlcNAc_site_dataset.gz
 Phosphorylation_site_dataset.gz Phosphosite_PTM_seq.fasta.gz
@@ -26,14 +26,14 @@ Phosphosite_seq.fasta.gz Regulatory_sites.gz Sumoylation_site_dataset.gz
 Ubiquitination_site_dataset.gz)
   
   ALL_FILES.each do |file|
-    PhosphoSitePlues.claim PhosphoSitePlues[".source"][file], :proc do |f|
+    PhosphoSitePlus.claim PhosphoSitePlus[".source"][file], :proc do |f|
       raise "Place #{file} from http://www.phosphosite.org/ at #{f}. Please consult license."
     end
   end
 
-  PhosphoSitePlues.claim PhosphoSitePlues.kinase_substrate, :proc do 
-    PhosphoSitePlues[".source/Kinase_Substrate_Dataset.gz"]
+  PhosphoSitePlus.claim PhosphoSitePlus.kinase_substrate, :proc do 
+    PhosphoSitePlus[".source/Kinase_Substrate_Dataset.gz"].produce
   end
 end
 
-iif PhosphoSitePlues.data.produce.find if __FILE__ == $0
+iif PhosphoSitePlus.data.produce.find if __FILE__ == $0

data/lib/rbbt/sources/pubmed.rb

@@ -228,26 +228,27 @@ module PubMed
       result = {}
       values = []
       chunks = Misc.divide(ids, (ids.length / 20) + 1)
-      bar = Log::ProgressBar.new_bar(chunks.length, :desc => "Downloading articles from PubMed")
-      chunks.each do |list|
-        begin
-          Misc.try3times do
-            url = "http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi" 
-
-            postdata = "db=pubmed&retmode=xml&id=#{list* ","}"
-            xml = TmpFile.with_file(postdata) do |postfile|
-              #Open.read(url, :quiet => true, :nocache => true, :nice => @@pubmed_lag, :nice_key => "PubMed", "--post-file=" => postfile)
-              Open.read(url+'?'+postdata, :quiet => true, :nocache => true, :nice => @@pubmed_lag, :nice_key => "PubMed", "--__post-file=" => postfile)
+      Log::ProgressBar.with_bar(chunks.length, :desc => "Downloading articles from PubMed") do |bar|
+        chunks.each do |list|
+          begin
+            Misc.try3times do
+              url = "http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi" 
+
+              postdata = "db=pubmed&retmode=xml&id=#{list* ","}"
+              xml = TmpFile.with_file(postdata) do |postfile|
+                #Open.read(url, :quiet => true, :nocache => true, :nice => @@pubmed_lag, :nice_key => "PubMed", "--post-file=" => postfile)
+                Open.read(url+'?'+postdata, :quiet => true, :nocache => true, :nice => @@pubmed_lag, :nice_key => "PubMed", "--__post-file=" => postfile)
+              end
+
+              values += xml.scan(/(<PubmedArticle>.*?<\/PubmedArticle>)/smu).flatten
             end
-
-            values += xml.scan(/(<PubmedArticle>.*?<\/PubmedArticle>)/smu).flatten
+          rescue Aborted
+            raise $!
+          rescue Exception
+            Log.exception $!
+          ensure
+            bar.tick
           end
-        rescue Aborted
-          raise $!
-        rescue Exception
-          Log.exception $!
-        ensure
-          bar.tick
         end
       end
 

data/lib/rbbt/sources/signor.rb

@@ -100,7 +100,42 @@ module Signor
 
     Misc.collapse_stream dumper.stream
   end
+
+  Signor.claim Signor.phospho_sites, :proc do
+    uni2name = UniProt.identifiers.Hsa.index :target => "Associated Gene Name", :fields => ["UniProt/SwissProt Accession"], :persist => true
+
+    dumper = TSV::Dumper.new :key_field => "Phosphosite", :fields => ["Effect"], :type => :flat, :organism => Signor.organism
+    dumper.init
+
+    TSV.traverse Signor.protein_protein, :into => dumper, :bar => true do |source, values|
+      res = []
+      res.extend MultipleResult
+      Misc.zip_fields(values).each do |target, effect, mechanism, residue|
+        kinase = case mechanism
+                 when "phosphorylation" 
+                   true
+                 when "dephosphorylation" 
+                   false
+                 else
+                   next
+                 end
+        name = uni2name[target]
+        next if name.nil?
+        next if residue.nil? or residue.empty?
+        site = [name, residue] * ":"
+        positive = effect.include? "up-regulates"
+
+        activates = kinase && positive || (!kinase && !positive)
+
+        res << [site, activates ? "Activates" : "Deactivates"]
+      end
+      res
+    end
+
+    TSV.collapse_stream(dumper)
+  end
 end
 
-iif Signor.tf_tg.produce(true).find if __FILE__ == $0
+iif Signor.tf_tg.produce.find if __FILE__ == $0
+iif Signor.phospho_sites.produce(true).find if __FILE__ == $0
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-sources
 version: !ruby/object:Gem::Version
-  version: 3.1.29
+  version: 3.1.30
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2018-01-15 00:00:00.000000000 Z
+date: 2018-05-14 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rbbt-util
@@ -90,6 +90,7 @@ files:
 - etc/biomart/missing_in_archive
 - etc/organisms
 - lib/rbbt/sources/CASCADE.rb
+- lib/rbbt/sources/COREAD_phospho_proteome.rb
 - lib/rbbt/sources/COSTART.rb
 - lib/rbbt/sources/CTCAE.rb
 - lib/rbbt/sources/GTRD.rb
@@ -100,6 +101,7 @@ files:
 - lib/rbbt/sources/PRO.rb
 - lib/rbbt/sources/PSI_MI.rb
 - lib/rbbt/sources/STITCH.rb
+- lib/rbbt/sources/array_express.rb
 - lib/rbbt/sources/barcode.rb
 - lib/rbbt/sources/bibtex.rb
 - lib/rbbt/sources/biomart.rb