1 """GeneOntology id retriver using Entrez and GenBank id --[cyppi]"""
   2 
   3 from Bio import GenBank, Fasta
   4 
   5 class getGoid:
   6     """ Get the goid from gid.
   7         this class fetch the Genbank document by given gid
   8         then parse the document and get goid(s)
   9         if not in the document or if there is some errer, 
  10         this class will return just empty list.
  11 
  12         usage :
  13         go = getGoid()
  14         golist = go.getgoid(gid='some gid', data='protein')
  15         the data is set 'protein' as a default,
  16         if the gid is protein, then just put gid without anynotation
  17         like 
  18         golist = go.getgoid('some gid')
  19         and if the gid related withe the dna
  20         data should be noted like
  21         golist = go.getgoid(gid='some gid', data='nucleotide')
  22 
  23 
  24     """
  25     fpser =''
  26     ncbi = ''
  27     ncbj = ''
  28     golist = []
  29     def __init__(self):
  30         self.fpser = GenBank.FeatureParser()
  31         self.ncbi = GenBank.NCBIDictionary(database = 'protein',format='genbank',parser=self.fpser)
  32         self.ncbj = GenBank.NCBIDictionary(database = 'nucleotide',format='genbank', parser = self.fpser)
  33 
  34     def getProteinGo(self,gid):
  35         try:    
  36             data = self.ncbi.get(gid)
  37             for item in data.features:
  38                 if item.qualifiers != '' and item.qualifiers.has_key('note'):
  39                     self.getgoidfromnote(item.qualifiers['note'])
  40         except :
  41             self.golist = []
  42 
  43     def getDNAGo(self, gid):
  44         try:
  45             data = self.ncbj.get(gid)
  46             for item in data.features:
  47                 if item.qualifiers != '' and item.qualifiers.has_key('note'):
  48                     self.getgoidfromnote(item.qualifiers['note'])
  49         except: 
  50             self.golist = []
  51 
  52     def getgoidfromnote(self, note):
  53         self.golist = []
  54         for item in note:
  55             item = item.replace('\n', ' ' )
  56             nlist = item.split('goid')
  57             if len(nlist) > 0:
  58                 for item in nlist[1:]:
  59                     if item[1:8] != '       ':
  60         #                print item[1:8]
  61                         self.golist.append(item[1:8])
  62 
  63     def getgoid(self,gid,data='protein'):
  64         if data== 'protein':
  65             self.getProteinGo(gid)
  66         elif data =='nucleotide':
  67             self.getDNAGo(gid)
  68         return self.golist
  69 
  70 
  71 if __name__ =='__main__':
  72     go  = getGoid()
  73     go.getgoid('6755536')
  74     for item in go.golist:
  75         print item