qpalma/parsers.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3
   4 class ReadParser:
   5    """
   6    A base class for the Solexa reads parsers.
   7    """
   8
   9    def __init__(self,filename):
  10       self.fh = open(filename)
  11
  12    def __iter__(self):
  13       return self
  14
  15    def next(self):
  16       pass
  17
  18
  19 class FilteredReadParser(ReadParser):
  20    """
  21    This class offers a parser for the reads that are created by the first
  22    filtering step performed to cut and join new reads.
  23    """
  24
  25    def __init__(self,filename):
  26       ReadParser.__init__(filename)
  27
  28    def parseLine(self,line):
  29       """
  30       We assume that a line has the following entires:
  31
  32       """
  33       return id,chr,pos,strand,mismatches,align_len,offset,alignment = line.split()
  34
  35    def next(self):
  36       for line in self.fh:
  37          line = line.strip()
  38           yield self.parseLine
  39
  40       raise StopIteration
  41
  42
  43 class RemappedReadParser(ReadParser):
  44    """
  45    This class offers a parser for the reads that are remapped by the vmatch
  46    utility.
  47    """
  48
  49    def __init__(self,filename):
  50       ReadParser.__init__(filename)
  51
  52    def parseLine(self,line):
  53       """
  54       We assume that a line has the following entires:
  55
  56       """
  57       return id,chr,pos,strand,mismatches,align_len,offset,alignment = line.split()
  58
  59    def next(self):
  60       for line in self.fh:
  61          line = line.strip()
  62           yield self.parseLine
  63
  64       raise StopIteration