bio 1.2.1 → 1.3.0

data/lib/bio/appl/blast/ddbj.rb

@@ -0,0 +1,142 @@
+#
+# = bio/appl/blast/ddbj.rb - Remote BLAST wrapper using DDBJ web service
+# 
+# Copyright::  Copyright (C) 2008       Naohisa Goto <ng@bioruby.org>
+# License::    The Ruby License
+#
+# $Id:$
+#
+
+require 'bio/appl/blast/remote'
+require 'bio/io/ddbjxml'
+
+module Bio::Blast::Remote
+
+  # Remote BLAST factory using DDBJ Web API for Biology
+  # (http://xml.nig.ac.jp/).
+  #
+  module DDBJ
+
+    # Creates a remote BLAST factory using DDBJ.
+    # Returns Bio::Blast object.
+    #
+    # Note for future improvement: In the future, it might return
+    # Bio::Blast::Remote::DDBJ or other object. 
+    #
+    def self.new(program, db, options = [])
+      Bio::Blast.new(program, db, options, 'ddbj')
+    end
+
+    # Information about DDBJ BLAST.
+    module Information
+
+      include Bio::Blast::Remote::Information
+
+      # (private) parse database information
+      def _parse_databases
+        if defined? @parse_databases
+          return nil if @parse_databases
+        end
+        drv = Bio::DDBJ::XML::Blast.new
+        str = drv.getSupportDatabaseList
+
+        databases = {}
+        dbdescs = {}
+        key = 'blastn'
+        prefix = ''
+        databases[key] ||= []
+        dbdescs[key] ||= {}
+        str.each_line do |line|
+          a = line.strip.split(/\s*\-\s*/, 2)
+          case a.size
+          when 1
+            prefix = a[0].to_s.strip
+            prefix += ': ' unless prefix.empty?
+            key = 'blastn'
+            next #each_line
+          when 0
+            prefix = ''
+            key = 'blastp'
+            databases[key] ||= []
+            dbdescs[key] ||= {}
+            next #each_line
+          end
+          name = a[0].to_s.strip.freeze
+          desc = (prefix + a[1].to_s.strip).freeze
+          databases[key].push name
+          dbdescs[key][name] = desc
+        end
+
+        databases['blastp'] ||= []
+        dbdescs['blastp'] ||= []
+
+        databases['blastn'].freeze
+        databases['blastp'].freeze
+
+        databases['blastx']  = databases['blastp']
+        dbdescs['blastx']    = dbdescs['blastp']
+        databases['tblastn'] = databases['blastn']
+        dbdescs['tblastn']   = dbdescs['blastn']
+        databases['tblastx'] = databases['blastn']
+        dbdescs['tblastx']   = dbdescs['blastn']
+
+        @databases = databases
+        @database_descriptions = dbdescs
+        @parse_databases = true
+        true
+      end
+      private :_parse_databases
+
+    end #module Information
+
+    extend Information
+
+    # executes BLAST and returns result as a string
+    def exec_ddbj(query)
+      options = make_command_line_options
+      opt = Bio::Blast::NCBIOptions.new(options)
+
+      # SOAP objects are cached
+      @ddbj_remote_blast ||= Bio::DDBJ::XML::Blast.new
+      #@ddbj_request_manager ||= Bio::DDBJ::XML::RequestManager.new
+      # always use REST version to prevent warning messages
+      @ddbj_request_manager ||= Bio::DDBJ::XML::RequestManager::REST.new
+
+      program = opt.delete('-p')
+      db = opt.delete('-d')
+      optstr = Bio::Command.make_command_line_unix(opt.options)
+
+      # using searchParamAsync 
+      qid = @ddbj_remote_blast.searchParamAsync(program, db, query, optstr)
+      @output = qid
+
+      sleeptime = 2
+      flag = true
+      while flag
+        if $VERBOSE then
+          $stderr.puts "DDBJ BLAST: ID: #{qid} -- waitng #{sleeptime} sec."
+        end
+        sleep(sleeptime)
+
+        result = @ddbj_request_manager.getAsyncResult(qid)
+        case result.to_s
+        when /The search and analysis service by WWW is very busy now/
+          raise result.to_s.strip + '(Alternatively, wrong options may be given.)'
+        when /Your job has not completed yet/
+          sleeptime = 5
+        else
+          flag = false
+        end
+      end while flag
+
+      @output = result
+      return @output
+    end
+
+  end #module DDBJ
+
+  # for lazy load DDBJ module
+  Ddbj = DDBJ
+
+end #module Bio::Blast::Remote
+

data/lib/bio/appl/blast/format0.rb

@@ -4,7 +4,7 @@
 # Copyright::  Copyright (C) 2003-2006 GOTO Naohisa <ng@bioruby.org>
 # License::    The Ruby License
 #
-# $Id: format0.rb,v 1.25 2007/12/27 17:28:57 ngoto Exp $
+# $Id:$
 #
 # == Description
 #
@@ -264,8 +264,8 @@ module Bio
               begin
                 q << sc.scan(/.*/)
                 sc.skip(/\s*^ ?/)
-              end until !sc.rest or r = sc.skip(/ *\( *(\d+) *letters *\)\s*\z/)
-              @query_len = sc[1].to_i if r
+              end until !sc.rest or r = sc.skip(/ *\( *([\,\d]+) *letters *\)\s*\z/)
+              @query_len = sc[1].delete(',').to_i if r
               @query_def = q.join(' ')
             end
           end
@@ -274,7 +274,7 @@ module Bio
         # Parses the first line of the BLAST result.
         def format0_parse_header
           unless defined?(@program)
-            if /(\w+) +([\w\-\.\d]+) *\[ *([\-\.\w]+) *\] *(\[.+\])?/ =~ @f0header.to_s
+            if /([\-\w]+) +([\w\-\.\d]+) *\[ *([\-\.\w]+) *\] *(\[.+\])?/ =~ @f0header.to_s
               @program = $1
               @version = "#{$1} #{$2} [#{$3}]"
               @version_number = $2
@@ -292,7 +292,17 @@ module Bio
             @f0references.push data.shift
           end
           @f0query = data.shift
+          # In special case, a void line is inserted after query name.
+          if data[0] and /\A +\( *([\,\d]+) *letters *\)\s*\z/ =~ data[0] then
+            @f0query.concat "\n"
+            @f0query.concat data.shift
+          end
           @f0database = data.shift
+          # In special case, a void line is inserted after database name.
+          if data[0] and /\A +[\d\,]+ +sequences\; +[\d\,]+ total +letters\s*\z/ =~ data[0] then
+            @f0database.concat "\n"
+            @f0database.concat data.shift
+          end
         end
 
         # Splits the statistical parameters.
@@ -340,7 +350,7 @@ module Bio
               sc = StringScanner.new(str)
               sc.skip(/\s*/)
               while sc.rest?
-                if sc.match?(/Number of sequences better than +([e\-\.\d]+) *\: *(.+)/) then
+                if sc.match?(/Number of sequences better than +([e\+\-\.\d]+) *\: *(.+)/) then
                   ev = sc[1]
                   ev = '1' + ev if ev[0] == ?e
                   @expect = ev.to_f
@@ -364,7 +374,7 @@ module Bio
               parse_colon_separated_params(@hash, @f0params)
               #p @hash
               if val = @hash['Matrix'] then
-                if /blastn *matrix *\: *([e\-\.\d]+) +([e\-\.\d]+)/ =~ val then
+                if /blastn *matrix *\: *([e\+\-\.\d]+) +([e\+\-\.\d]+)/ =~ val then
                   @matrix = 'blastn'
                   @sc_match    = $1.to_i
                   @sc_mismatch = $2.to_i 
@@ -373,16 +383,16 @@ module Bio
                 end
               end
               if val = @hash['Gap Penalties'] then
-                if /Existence\: *([e\-\.\d]+)/ =~ val then
+                if /Existence\: *([e\+\-\.\d]+)/ =~ val then
                   @gap_open = $1.to_i
                 end
-                if /Extension\: *([e\-\.\d]+)/ =~ val then
+                if /Extension\: *([e\+\-\.\d]+)/ =~ val then
                   @gap_extend = $1.to_i
                 end
               end
               #@db_num = @hash['Number of Sequences'] unless defined?(@db_num)
               #@db_len = @hash['length of database']  unless defined?(@db_len)
-              if val = @hash['effective length of database'] then
+              if val = @hash['effective search space'] then
                 @eff_space = val.tr(',', '').to_i
               end
               @parse_params = true
@@ -529,7 +539,7 @@ module Bio
                   @hits << Hit.new(data)
                   r = data.first
                   break unless r
-                  if /^Significant alignments for pattern/ =~ r
+                  while /^Significant alignments for pattern/ =~ r
                     data.shift
                     r = data.first
                   end
@@ -584,9 +594,9 @@ module Bio
               @pattern_positions = []
               @f0message.each do |r|
                 sc = StringScanner.new(r)
-                if sc.skip_until(/^ *pattern +(.+)$/) then
+                if sc.skip_until(/^ *pattern +([^\s]+)/) then
                   @pattern = sc[1] unless @pattern
-                  sc.skip_until(/^ at position +(\d+)/)
+                  sc.skip_until(/(?:^ *| +)at position +(\d+) +of +query +sequence/)
                   @pattern_positions << sc[1].to_i
                 end
               end
@@ -711,19 +721,19 @@ module Bio
                   sc.skip(/ */)
                 end
                 sc.skip(/\s*/)
-                while r = sc.scan(/[e\.\-\d]+/)
+                while r = sc.scan(/[e\+\-\.\d]+/)
                   #p r
                   h[s0.shift] = r
                   sc.skip(/ */)
                 end
                 if gapped then
-                  @gapped_lambda = h['Lambda']
-                  @gapped_kappa = h['K']
-                  @gapped_entropy = h['H']
+                  @gapped_lambda = (v = h['Lambda']) ? v.to_f : nil
+                  @gapped_kappa = (v = h['K']) ? v.to_f : nil
+                  @gapped_entropy = (v = h['H']) ? v.to_f : nil
                 else
-                  @lambda = h['Lambda']
-                  @kappa = h['K']
-                  @entropy = h['H']
+                  @lambda = (v = h['Lambda']) ? v.to_f : nil
+                  @kappa = (v = h['K']) ? v.to_f : nil
+                  @entropy = (v = h['H']) ? v.to_f : nil
                 end
               end #each
               @parse_stat = true
@@ -861,7 +871,7 @@ module Bio
                 d << sc.scan(/.*/)
                 sc.skip(/\s*/)
               end until !sc.rest? or r = sc.skip(/ *Length *\= *([\,\d]+)\s*\z/)
-              @len = (r ? sc[1].to_i : nil)
+              @len = (r ? sc[1].delete(',').to_i : nil)
               @definition = d.join(" ")
               @parse_hitname = true
             end
@@ -968,11 +978,11 @@ module Bio
               sc = StringScanner.new(@f0score)
               while sc.rest?
                 sc.skip(/\s*/)
-                if sc.skip(/Expect(?:\(\d\))? *\= *([e\-\.\d]+)/) then
+                if sc.skip(/Expect(?:\(\d+\))? *\= *([e\+\-\.\d]+)/) then
                   ev = sc[1].to_s
                   ev = '1' + ev if ev[0] == ?e
                   @evalue = ev.to_f
-                elsif sc.skip(/Score *\= *([e\-\.\d]+) *bits *\( *([e\-\.\d]+) *\)/) then
+                elsif sc.skip(/Score *\= *([e\+\-\.\d]+) *bits *\( *([e\+\-\.\d]+) *\)/) then
                   bs = sc[1]
                   bs = '1' + bs if bs[0] == ?e
                   @bit_score = bs.to_f
@@ -1016,19 +1026,19 @@ module Bio
                   if sc[2] then
                     @hit_frame = sc[3].to_i
                   end
-                elsif sc.skip(/Score *\= *([e\-\.\d]+) +\(([e\-\.\d]+) *bits *\)/) then
+                elsif sc.skip(/Score *\= *([e\+\-\.\d]+) +\(([e\+\-\.\d]+) *bits *\)/) then
                   #WU-BLAST
                   @score = sc[1].to_i
                   bs = sc[2]
                   bs = '1' + bs if bs[0] == ?e
                   @bit_score = bs.to_f
-                elsif sc.skip(/P *\= * ([e\-\.\d]+)/) then
+                elsif sc.skip(/P *\= * ([e\+\-\.\d]+)/) then
                   #WU-BLAST
                   @p_sum_n = nil
                   pv = sc[1]
                   pv = '1' + pv if pv[0] == ?e
                   @pvalue = pv.to_f
-                elsif sc.skip(/Sum +P *\( *(\d+) *\) *\= *([e\-\.\d]+)/) then
+                elsif sc.skip(/Sum +P *\( *(\d+) *\) *\= *([e\+\-\.\d]+)/) then
                   #WU-BLAST
                   @p_sum_n = sc[1].to_i
                   pv = sc[2]

data/lib/bio/appl/blast/format8.rb

@@ -4,7 +4,7 @@
 # Copyright::  Copyright (C) 2002, 2003, 2007 Toshiaki Katayama <k@bioruby.org>
 # License::    The Ruby License
 #
-# $Id: format8.rb,v 1.8 2007/12/14 16:15:20 k Exp $
+# $Id:$
 #
 # == Note
 #
@@ -27,7 +27,7 @@ module Bio
         hit_num = 1
         hsp_num = 1
         hit = ''
-        data.each do |line|
+        data.each_line do |line|
           ary = line.chomp.split("\t")
           query_id, target_id, hsp = tab_parse_hsp(ary)
           if query_prev != query_id or target_prev != target_id

data/lib/bio/appl/blast/genomenet.rb

@@ -0,0 +1,263 @@
+#
+# = bio/appl/blast/genomenet.rb - Remote BLAST wrapper using GenomeNet
+# 
+# Copyright::  Copyright (C) 2001,2008  Mitsuteru C. Nakao <n@bioruby.org>
+# Copyright::  Copyright (C) 2002,2003  Toshiaki Katayama <k@bioruby.org>
+# Copyright::  Copyright (C) 2006       Jan Aerts <jan.aerts@bbsrc.ac.uk>
+# Copyright::  Copyright (C) 2008       Naohisa Goto <ng@bioruby.org>
+# License::    The Ruby License
+#
+# $Id:$
+#
+
+require 'net/http'
+require 'uri'
+require 'bio/command'
+require 'shellwords'
+require 'bio/appl/blast/remote'
+
+module Bio::Blast::Remote
+
+  # == Description
+  # 
+  # The Bio::Blast::Remote::GenomeNet class contains methods for running
+  # remote BLAST searches on GenomeNet (http://blast.genome.jp/).
+  #
+  # == Usage
+  #
+  #   require 'bio'
+  #   
+  #   # To run an actual BLAST analysis:
+  #   #   1. create a BLAST factory
+  #   blast_factory = Bio::Blast.remote('blastp', 'nr-aa',
+  #                                     '-e 0.0001', 'genomenet')
+  #   #or:
+  #   blast_factory = Bio::Blast::Remote.genomenet('blastp', 'nr-aa',
+  #                                                '-e 0.0001')
+  #
+  #   #   2. run the actual BLAST by querying the factory
+  #   report = blast_factory.query(sequence_text)
+  #
+  #   # Then, to parse the report, see Bio::Blast::Report
+  #
+  # === Available databases for Bio::Blast::Remote::GenomeNet
+  #
+  # Up-to-date available databases can be obtained by using
+  # Bio::Blast::Remote::GenomeNet.databases(program).
+  # Short descriptions of databases
+  #
+  #  ----------+-------+---------------------------------------------------
+  #   program  | query | db (supported in GenomeNet)
+  #  ----------+-------+---------------------------------------------------
+  #   blastp   | AA    | nr-aa, genes, vgenes.pep, swissprot, swissprot-upd,
+  #  ----------+-------+ pir, prf, pdbstr
+  #   blastx   | NA    | 
+  #  ----------+-------+---------------------------------------------------
+  #   blastn   | NA    | nr-nt, genbank-nonst, gbnonst-upd, dbest, dbgss,
+  #  ----------+-------+ htgs, dbsts, embl-nonst, embnonst-upd, epd,
+  #   tblastn  | AA    | genes-nt, genome, vgenes.nuc
+  #  ----------+-------+---------------------------------------------------
+  #
+  # == See also
+  #
+  # * Bio::Blast
+  # * Bio::Blast::Report
+  # * Bio::Blast::Report::Hit
+  # * Bio::Blast::Report::Hsp
+  #
+  # == References
+  # 
+  # * http://www.ncbi.nlm.nih.gov/blast/
+  # * http://www.ncbi.nlm.nih.gov/Education/BLASTinfo/similarity.html
+  # * http://blast.genome.jp/ideas/ideas.html#blast
+  #
+  module GenomeNet
+
+    Host = "blast.genome.jp".freeze
+
+    # Creates a remote BLAST factory using GenomeNet.
+    # Returns Bio::Blast object.
+    #
+    # Note for future improvement: In the future, it might return
+    # Bio::Blast::Remote::GenomeNet or other object. 
+    #
+    def self.new(program, db, options = [])
+      Bio::Blast.new(program, db, options, 'genomenet')
+    end
+
+    # Information for GenomeNet BLAST search.
+    module Information
+
+      include Bio::Blast::Remote::Information
+
+      # gets information from remote host and parses database information
+      def _parse_databases
+        if defined? @parse_databases
+          return nil if @parse_databases
+        end
+        databases = {}
+        dbdescs = {}
+        key = nil
+        host = Bio::Blast::Remote::Genomenet::Host
+        http = Bio::Command.new_http(host)
+        result = http.get('/')
+        #p result.body
+        result.body.each_line do |line|
+          case line
+          when /\"set\_dbtype\(this\.form\,\'(prot|nucl)\'\)\"/
+            key = $1
+            databases[key] ||= []
+            dbdescs[key] ||= {}
+          when /\<input *type\=\"radio\" *name\=\"dbname\" *value\=\"([^\"]+)\"[^\>]*\>([^\<\>]+)/
+            db = $1.freeze
+            desc = $2.strip.freeze
+            databases[key].push db
+            dbdescs[key][db] = desc
+          end
+        end
+
+        # mine-aa and mine-nt should be removed
+        [ 'prot', 'nucl' ].each do |mol|
+          ary  = databases[mol] || []
+          hash = dbdescs[mol] || {}
+          [ 'mine-aa', 'mine-nt' ].each do |k|
+            ary.delete(k)
+            hash.delete(k)
+          end
+          databases[mol] = ary.freeze
+          dbdescs[mol] = hash
+        end
+
+        [ databases, dbdescs ].each do |h|
+          prot = h['prot']
+          nucl = h['nucl']
+          h.delete('prot')
+          h.delete('nucl')
+          h['blastp'] = prot
+          h['blastx'] = prot
+          h['blastn']  = nucl
+          h['tblastn'] = nucl
+          h['tblastx'] = nucl
+        end
+
+        @databases = databases
+        @database_descriptions = dbdescs
+        @parse_databases = true
+        true
+      end
+      private :_parse_databases
+
+    end #module Information
+
+    extend Information
+
+    private
+
+    # executes BLAST and returns result as a string
+    def exec_genomenet(query)
+      host = Host
+      #host = "blast.genome.jp"
+      #path = "/sit-bin/nph-blast"
+      path = "/sit-bin/blast" #2005.08.12
+
+      options = make_command_line_options
+      opt = Bio::Blast::NCBIOptions.new(options)
+
+      program = opt.delete('-p')
+      db = opt.delete('-d')
+
+      matrix = opt.delete('-M') || 'blosum62'
+      filter = opt.delete('-F') || 'T'
+
+      opt_V = opt.delete('-V') || 500 # default value for GenomeNet
+      opt_B = opt.delete('-B') || 250 # default value for GenomeNet
+
+      # format, not for form parameters, but included in option string
+      opt_m = opt.get('-m') || '7' # default of BioRuby GenomeNet factory
+      opt.set('-m', opt_m)
+
+      optstr = Bio::Command.make_command_line_unix(opt.options)
+
+      form = {
+        'style'          => 'raw',
+        'prog'           => program,
+        'dbname'         => db,
+        'sequence'       => query,
+        'other_param'    => optstr,
+        'matrix'         => matrix,
+        'filter'         => filter,
+        'V_value'        => opt_V, 
+        'B_value'        => opt_B, 
+        'alignment_view' => 0,
+      }
+
+      form.keys.each do |k|
+        form.delete(k) unless form[k]
+      end
+
+      begin
+        http = Bio::Command.new_http(host)
+        http.open_timeout = 300
+        http.read_timeout = 600
+        result = Bio::Command.http_post_form(http, path, form)
+        @output = result.body
+
+        # workaround 2008.8.13
+        if result.code == '302' then
+          newuri = URI.parse(result['location'])
+          newpath = newuri.path
+          result = http.get(newpath)
+          @output = result.body
+          # waiting for BLAST finished
+          while /Your job ID is/ =~ @output and
+              /Your result will be displayed here\<br\>/ =~ @output
+            if /This page will be reloaded automatically in\s*((\d+)\s*min\.)?\s*(\d+)\s*sec\./ =~ @output then
+              reloadtime = $2.to_i * 60 + $3.to_i
+              reloadtime = 300 if reloadtime > 300
+              reloadtime = 1 if reloadtime < 1
+            else
+              reloadtime = 5
+            end
+            if $VERBOSE then
+              $stderr.puts "waiting #{reloadtime} sec to reload #{newuri.to_s}"
+            end
+            sleep(reloadtime)
+            result = http.get(newpath)
+            @output = result.body
+          end
+        end
+
+        # workaround 2005.08.12
+        if /\<A +HREF=\"(http\:\/\/blast\.genome\.jp(\/tmp\/[^\"]+))\"\>Show all result\<\/A\>/i =~ @output.to_s then
+          result = http.get($2)
+          @output = result.body
+          txt = @output.to_s.split(/\<pre\>/)[1]
+          raise 'cannot understand response' unless txt
+          txt.sub!(/\<\/pre\>.*\z/m, '')
+          txt.sub!(/.*^ \-{20,}\s*/m, '')
+          @output = txt.gsub(/\&lt\;/, '<')
+        else
+          raise 'cannot understand response'
+        end
+      end
+
+      # for -m 0 (NCBI BLAST default) output, html tags are removed.
+      if opt_m.to_i == 0 then
+        #@output_bak = @output
+        txt = @output.gsub(/^\s*\<img +src\=\"\/Fig\/arrow\_top\.gif\"\>.+$\r?\n/, '')
+        txt.gsub!(/^.+\<\/form\>$/, '')
+        txt.gsub!(/^\<form *method\=\"POST\" name\=\"clust\_check\"\>.+$\r?\n/, '')
+        txt.gsub!(/\<[^\>\<]+\>/m, '')
+        @output = txt
+      end
+
+      return @output
+    end
+
+  end # class GenomeNet
+
+  # alias for lazy load
+  Genomenet = GenomeNet
+
+end # module Bio::Blast::Remote
+