pysam/Pileup.py

   1 '''Tools for working with files in the samtools pileup -c format.'''
   2 import collections
   3 import pysam
   4
   5 PileupSubstitution = collections.namedtuple( "PileupSubstitution",
   6                                     " ".join( (\
   7             "chromosome",
   8             "position",
   9             "reference_base",
  10             "consensus_base",
  11             "consensus_quality",
  12             "snp_quality",
  13             "rms_mapping_quality",
  14             "coverage",
  15             "read_bases",
  16             "base_qualities" ) ) )
  17
  18 PileupIndel = collections.namedtuple( "PileupIndel",
  19                                       " ".join( (\
  20             "chromosome",
  21             "position",
  22             "reference_base",
  23             "genotype",
  24             "consensus_quality",
  25             "snp_quality",
  26             "rms_mapping_quality",
  27             "coverage",
  28             "first_allelle",
  29             "second_allele",
  30             "reads_first",
  31             "reads_second",
  32             "reads_diff" ) ) )
  33
  34 def iterate( infile ):
  35     '''iterate over ``samtools pileup -c`` formatted file.
  36
  37     *infile* can be any iterator over a lines.
  38
  39     The function yields named tuples of the type :class:`pysam.Pileup.PileupSubstitution`
  40     or :class:`pysam.Pileup.PileupIndel`.
  41
  42     .. note::
  43        The parser converts to 0-based coordinates
  44     '''
  45
  46     conv_subst = (str,lambda x: int(x)-1,str,str,int,int,int,int,str,str)
  47     conv_indel = (str,lambda x: int(x)-1,str,str,int,int,int,int,str,str,int,int,int)
  48
  49     for line in infile:
  50         d = line[:-1].split()
  51         if d[2] == "*":
  52             try:
  53                 yield PileupIndel( *[x(y) for x,y in zip(conv_indel,d) ] )
  54             except TypeError:
  55                 raise pysam.SamtoolsError( "parsing error in line: `%s`" % line)
  56         else:
  57             try:
  58                 yield PileupSubstitution( *[x(y) for x,y in zip(conv_subst,d) ] )
  59             except TypeError:
  60                 raise pysam.SamtoolsError( "parsing error in line: `%s`" % line)