#! /usr/bin/env python
# -*- coding: cp936 -*-
from operator import itemgetter
# ²âÊÔ»ùÒò×Ö·û´®
s = open(r'gene.txt','r').read()
# µÃµ½Ã¿¸ö»ùÒò×Ö·û´®
m = ['>'+x for x in s.split('>')[1:]]
# Éú³É »ùÒò×Ö·û´®:°üº¬2¸ö×Ö·û´®ÊýÄ¿Ö®ºÍ µÄ×Öµä
d = {}
for c in m:
        d[c] = c.upper().count('ATCG')+c.upper().count('GCTAT')
# °´ÕÕÊýÄ¿Ö®ºÍÅÅÐò,´Ó´óµ½Ð¡,µÃµ½Ò»¸ölist of tuple
d = sorted(d.iteritems(), key=itemgetter(1), reverse=True)
# ¹¹Ôì½á¹û×Ö·û´®
ss = ''
for ge in d:
        # »ùÒòID
        refGene = ge[0][ge[0].index('_refGene_')+len('_refGene_'):ge[0].index('range')]
        # Ëùº¬×Ö·û´®ÊýÄ¿Ö®ºÍ
        strNum = ge[1]
        # ¼î»ùÊýÄ¿
        baseNum = sum([x.upper() in 'ATCG' and 1 or 0 for x in ge[0][ge[0].index('none')+len('none'):]])
        ss += refGene+'\t'+str(strNum)+'\t'+str(baseNum)+'\n'
# Êä³ö½á¹ûµ½Îļþ
f = open(r'data.txt','w')
f.write(ss)
f.close()
print 'Done'

NR_030165         2        2000
NM_052933         2        2000
NM_005989         0        2000
NM_001190906         0        2000

2Â¥: Originally posted by libralibra at 2012-12-05 21:05:29
×Ö·û´®½âÎöÓýű¾ÓïÑÔÓ¦¸ÃÊÇ×îˬµÄ,ÏÂÃæÊǸöpython½Å±¾,»ùÒò×Ö·û´®±£´æÔÚgene.txt,¸úÕâ¸ö.pyÎļþ·ÅÔÚͬһ¸öÎļþ¼ÐÏÂ,ÔËÐÐÍêºó»áÉú³Édata.txt,ÀïÃæÊÇ»ùÒòid,×Ö·û´®ÊýÄ¿ºÍ¼î»ùÊýÄ¿3ÁÐÊý¾Ý,ÓÃtab·Ö¸î
ÕâÑùµÄºÃ´¦ÊÇÖ±½Ó ...