uniprot.rb in rbbt-sources-2.1.0

- old
+ new

@@ -1,5 +1,6 @@
+require 'rbbt'
 require 'rbbt/util/open'
 require 'rbbt/resource'
 require 'rbbt/sources/cath'
 require 'rbbt/sources/uniprot'
 
@@ -31,10 +32,11 @@
     tsv.to_s
   end
 
 
   UNIPROT_TEXT="http://www.uniprot.org/uniprot/[PROTEIN].txt"
+  UNIPROT_FASTA="http://www.uniprot.org/uniprot/[PROTEIN].fasta"
   def self.pdbs(protein)
     url = UNIPROT_TEXT.sub "[PROTEIN]", protein
     text = Open.read(url)
 
     pdb = {}
@@ -42,17 +44,72 @@
     text.split(/\n/).each{|l| 
       next unless l =~ /^DR\s+PDB; (.*)\./
       id, method, resolution, region = $1.split(";").collect{|v| v.strip}
       begin
         chains, start, eend = region.match(/(\w+)=(\d+)-(\d+)/).values_at(1,2,3)
+        start = start.to_i
+        eend = eend.to_i
+        start, eend = eend, start if start > eend
       rescue
         Log.warn("Error process Uniprot PDB line: #{line}")
         next
       end
-      pdb[id.downcase] = {:method => method, :resolution => resolution, :region => (start.to_i..eend.to_i), :chains => chains}
+      pdb[id.downcase] = {:method => method, :resolution => resolution, :region => (start..eend), :chains => chains}
     }
     pdb
   end
+
+  def self.sequence(protein)
+    url = UNIPROT_FASTA.sub "[PROTEIN]", protein
+    text = Open.read(url)
+
+    text.split(/\n/).select{|line| line !~ /^>/} * ""
+  end
+
+  def self.features(protein)
+    url = UNIPROT_TEXT.sub "[PROTEIN]", protein
+    text = Open.read(url)
+
+    text = text.split(/\n/).select{|line| line =~ /^FT/} * "\n"
+
+    parts = text.split(/^(FT   \w+)/)
+    parts.shift
+
+    features = []
+
+    type = nil
+    parts.each do |part|
+      parts
+      if part[0..1] == "FT"
+        type = part.gsub(/FT\s+/,'')
+        next
+      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 = $1, $2
+        description = nil
+      else
+        Log.debug "Value not understood: #{ value }"
+      end
+
+
+      feature = {
+        :type => type,
+        :start => start.to_i, 
+        :end => eend.to_i, 
+        :description => description,
+      }
+
+      features << feature
+    end
+
+    features
+  end
+
 
   def self.variants(protein)
     url = UNIPROT_TEXT.sub "[PROTEIN]", protein
     text = Open.read(url)
 