1 """This parser is working when using "-v -cds" options"""
   2 class GenscanReport:
   3     def __init__(self):
   4         self.genes = list()
   5     def addGene(self, gene):
   6         self.genes.append(gene)
   7 
   8 class Gene:
   9     def __init__(self, no):
  10         self.no = no
  11         self.exons = list()
  12         self.cds = ''
  13         self.peptide = ''
  14     def addExon(self, exon):
  15         self.exons.append(exon)
  16 
  17     def setCds(self, fasta):
  18         self.cds = fasta
  19     def setPeptide(self, fasta):
  20         self.peptide = fasta
  21 
  22 class Exon:
  23     def __init__(self, no, type, strand, begin, end):
  24         self.no = no
  25         self.type = type
  26         self.strand = strand
  27         self.begin = int(begin)
  28         self.end = int(end)
  29 
  30 class FastaIterator:
  31     def __init__(self, ifile):
  32         self.ifile = ifile
  33         self.g = self.getGenerator()
  34     def getGenerator(self):
  35         while True:
  36             line = self.ifile.next().strip()
  37             if line:
  38                 lines = [line+'\n']
  39                 break
  40         for line in self.ifile:
  41             if line.startswith('Explanation'):
  42                 yield ''.join(lines)
  43                 break
  44             if line.startswith('>'):
  45                 yield ''.join(lines)
  46                 lines = [line]
  47             else:
  48                 lines.append(line)
  49 
  50     def __iter__(self):
  51         return self.g
  52 
  53     def next(self):
  54         return self.g.next()
  55 
  56 def parseGenscan(handler):
  57     gr = GenscanReport()
  58     while True:
  59         line = handler.next()
  60         if line.startswith('-') or line is None: break
  61     i = 0
  62     gene = None
  63     for line in handler:
  64         line = line.strip()
  65         if line.startswith('P') or line.startswith('S') or line.startswith('NO'): 
  66             break
  67         if not line:
  68             if gene and gene.exons: gr.addGene(gene)
  69             i+=1
  70             gene = Gene(i)
  71             continue
  72         words=line.split()
  73         gene.addExon(Exon(words[0], words[1], words[2], words[3], words[4]))
  74 
  75     while True:
  76         line = handler.next()
  77         if line.startswith('Predicted coding') or line is None: break
  78 
  79     fi = FastaIterator(handler)
  80     for gene in gr.genes:
  81         gene.setPeptide(fi.next())
  82         gene.setCds(fi.next())
  83     return gr