Source Code for Module Bio.SeqIO.IgIO

  1  # Copyright 2008-2010 by Peter Cock.  All rights reserved. 
  2  # This code is part of the Biopython distribution and governed by its 
  3  # license.  Please see the LICENSE file that should have been included 
  4  # as part of this package. 
  5  # 
  6  # This module is for reading and writing IntelliGenetics format files as 
  7  # SeqRecord objects.  This file format appears to be the same as the MASE 
  8  # multiple sequence alignment format. 
  9   
 10  """Bio.SeqIO support for the "ig" (IntelliGenetics or MASE) file format. 
 11   
 12  You are expected to use this module via the Bio.SeqIO functions.""" 
 13   
 14  from Bio.Alphabet import single_letter_alphabet 
 15  from Bio.Seq import Seq 
 16  from Bio.SeqRecord import SeqRecord 
 17   
 18   
 19 -def IgIterator(handle, alphabet=single_letter_alphabet): 
 20      """Iterate over IntelliGenetics records (as SeqRecord objects). 
 21   
 22      handle - input file 
 23      alphabet - optional alphabet 
 24   
 25      The optional free format file header lines (which start with two 
 26      semi-colons) are ignored. 
 27   
 28      The free format commentary lines at the start of each record (which 
 29      start with a semi-colon) are recorded as a single string with embedded 
 30      new line characters in the SeqRecord's annotations dictionary under the 
 31      key 'comment'. 
 32      """ 
 33      #Skip any file header text before the first record (;; lines) 
 34      while True: 
 35          line = handle.readline() 
 36          if not line: 
 37              break  # Premature end of file, or just empty? 
 38          if not line.startswith(";;"): 
 39              break 
 40   
 41      while line: 
 42          #Now iterate over the records 
 43          if line[0] != ";": 
 44              raise ValueError( 
 45                  "Records should start with ';' and not:\n%s" % repr(line)) 
 46   
 47          #Try and agree with SeqRecord convention from the GenBank parser, 
 48          #(and followed in the SwissProt parser) which stores the comments 
 49          #as a long string with newlines under annotations key 'comment'. 
 50   
 51          #Note some examples use "; ..." and others ";..." 
 52          comment_lines = [] 
 53          while line.startswith(";"): 
 54              #TODO - Extract identifier from lines like "LOCUS\tB_SF2"? 
 55              comment_lines.append(line[1:].strip()) 
 56              line = handle.readline() 
 57          title = line.rstrip() 
 58   
 59          seq_lines = [] 
 60          while True: 
 61              line = handle.readline() 
 62              if not line: 
 63                  break 
 64              if line[0] == ";": 
 65                  break 
 66              #Remove trailing whitespace, and any internal spaces 
 67              seq_lines.append(line.rstrip().replace(" ", "")) 
 68          seq_str = "".join(seq_lines) 
 69          if seq_str.endswith("1"): 
 70              #Remove the optional terminator (digit one) 
 71              seq_str = seq_str[:-1] 
 72          if "1" in seq_str: 
 73              raise ValueError( 
 74                  "Potential terminator digit one found within sequence.") 
 75   
 76          #Return the record and then continue... 
 77          record = SeqRecord(Seq(seq_str, alphabet), 
 78                             id=title, name=title) 
 79          record.annotations['comment'] = "\n".join(comment_lines) 
 80          yield record 
 81   
 82      #We should be at the end of the file now 
 83      assert not line 
 84   
 85  if __name__ == "__main__": 
 86      print "Running quick self test" 
 87   
 88      import os 
 89      path = "../../Tests/IntelliGenetics/" 
 90      if os.path.isdir(path): 
 91          for filename in os.listdir(path): 
 92              if os.path.splitext(filename)[-1] == ".txt": 
 93                  print 
 94                  print filename 
 95                  print "-" * len(filename) 
 96                  handle = open(os.path.join(path, filename)) 
 97                  for record in IgIterator(handle): 
 98                      print record.id, len(record) 
 99                  handle.close() 
100          print "Done" 
101      else: 
102          print "Could not find input files" 
103