allpy: f744a22567fb repeats/test.py

allpy

repeats: tab-separated output

author	boris (netbook) <bnagaev@gmail.com>
date	Fri, 26 Nov 2010 18:38:00 +0300
parents	2a0ad666fa19
children	54e898f96d80

line source

1 from repeat_joiner import Repeat, RepeatJoiner, RepeatPair

2 import pprint

4 rj = RepeatJoiner()

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

6 line = line.strip()

7 if line:

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

9 if c1 == 'DNA_1':

10 continue # first line

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

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

13 from1 = int(from1)

14 to1 = int(to1) + 1

15 from2 = int(from2)

16 to2 = int(to2) + 1

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

20 rj.build_groups()

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

22 print "group\tchr\tchr_from\tchr_to\tgroup_from\tgroup_to\tori\tgroup_ori"

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

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

25 # pprint.pprint(repeat_group)

26 for repeat in repeat_group:

27 ori = 1 if repeat.ori else -1

28 group_ori = 1 if repeat.group_ori else -1

29 print "%i\t%s\t%i\t%i\t%i\t%i\t%i\t%i" % (i, repeat.chromosome, repeat.start,

30 repeat.end-1, repeat.group_start, repeat.group_end-1, ori, group_ori)