badgerBtab

#!/usr/bin/env ruby

require 'ostruct'
require 'optparse'
require 'rubygems'
require 'bio'
opt = OpenStruct.new

opt.top = false
ARGV.options do |opts|
  opts.banner << " blast-file [...blast-file...]"
  opts.on("-t", "--top","only show top hit") {opt.top = true}    
  begin
    opts.parse!
  rescue
    STDERR.puts $!.message
    STDERR.puts opts
    exit(1)
  end
end

ARGV.push(STDIN) if ARGV.empty?

ARGV.each do |blast|
  if (blast == STDIN)
    out = STDOUT
  else
    out = File.new(blast + ".btab", "w")
  end
  Bio::Blast::Default::Report.open(blast, "r").each do |query|
    fields = [""]*21
    fields[0] = query.query_def.split(" ").first
    fields[2] = query.query_len
    fields[3] = query.program
    fields[4] = query.db
    begin
      query.each do |hit|
        hsp = hit.hsps.first
        fields[5] = hit.definition.split(" ").first
        fields[6] = hit.query_start
        fields[7] = hit.query_end
        fields[8] = hit.target_start
        fields[9] = hit.target_end
        hitLen = (hit.target_start - hit.target_end).abs + 1
        fields[10] = ((hit.identity / hitLen.to_f)*1000).to_i/10.0
        fields[12] = hit.score
        fields[13] = hit.bit_score
        fields[15] = hit.definition
        fields[16] = hsp.query_frame
        fields[17] = hsp.query_strand
        fields[18] = query.db_len
        fields[19] = hit.evalue
        fields[20] = hit.evalue
        fields.each do |field|
          out.printf("%s\t", field.to_s)
        end
        out.printf("\n")
        break if opt.top
      end
    rescue
      STDERR.printf("Skipping query %s...\n", fields[0]) if (out != STDOUT)
    end
  end
  out.close
end

# This is a BioRuby remake. The original BTAB program is described here:
# Dubnick M. (1992) Btab--a Blast output parser. Comput Appl Biosci 8(6):601-2
# The parsed BLAST output is presented as single line HSP descriptions, with tab-delimited 
# fields in the following order:
#
# [0]      Query Sequence Name
# [2]      Query Sequence Length
# [3]      Search Method  --  Blast family application name
# [4]      Database Name
# [5]      Subject Sequence Name  --  Database entry name
# [6],[7]   Query Left End, Query Right End  --  The endpoints of the part
#                of the query sequence which Blast aligns with the subject sequence.
# [8],[9]  Subject Left End, Subject Right End  --  The endpoints of the part 
#                of the subject sequence which Blast aligns with the query sequence.
# [10]     Percent Identity  --  The fraction of residues which are absolute
#             matches between the query and subject sequence, expressed in
#             percent.  
# [12]    HSP score    
# [13]    Bits score
# [15]    Description  --  A freeform text field which contains the biological
#         description field from the database for the subject sequence.
# [16]    Query Frame (1, 2, 3, -1, -2, -3)
# [17]    Query Strand  --  Plus, Minus or null
# [18]    DB sequence length
# [19]    Expect -- expected value
# [20]    P-Value  --  Poisson ratio
#
# ** Note ** Intervening field positions which are not described are not currently supported.
# These remain to support compatibility with other existing btab implementations.