allpy: 2a0ad666fa19 repeats/test.py

allpy

repeats: documentation; doctest corrected

author	boris (netbook) <bnagaev@gmail.com>
date	Wed, 24 Nov 2010 14:57:24 +0300
parents	32d244bf868a
children	f744a22567fb

line source

1 from repeat_joiner import Repeat, RepeatJoiner, RepeatPair

2 import pprint

4 rj = RepeatJoiner()

5 print 'start adding repeats'

6 for line in open('Complete_pair_repeats.txt'):

7 line = line.strip()

8 if line:

9 c1, c2, from1, to1, from2, to2, ori1, ori2, l1, l2 = line.split()

10 if c1 == 'DNA_1':

11 continue # first line

12 ori1 = True if ori1 == '1' else False

13 ori2 = True if ori2 == '1' else False

14 from1 = int(from1)

15 to1 = int(to1) + 1

16 from2 = int(from2)

17 to2 = int(to2) + 1

19 RepeatPair([Repeat(rj, c1, from1, to1, ori1), Repeat(rj, c2, from2, to2, ori2)])

21 print 'all repeats have been added'

22 rj.build_groups()

23 rj.repeat_groups.sort(key=lambda g: len(g), reverse=True)

24 print 'full repeats were found'

25 for i, repeat_group in enumerate(rj.repeat_groups):

26 print 'Repeat group', i

27 repeat_group.sort(key=lambda r: min(r.group_start, r.group_end))

28 pprint.pprint(repeat_group)