view test/usecase2.py @ 666:d9a3e99ae19b
blocks3d/www: mark conservative columns inside blocks
Previously letters were marked if column is conservative over all
sequences of alignment. Currently it was changed to conservatism
over block including this letter.
Additional attribute storing conservatism of columns inside block
was added to block. This caused all colors of blocks to be changed.
| author |
boris <bnagaev@gmail.com> |
| date |
Fri, 01 Jul 2011 03:21:08 +0400 |
| parents |
bed32775625a |
| children |
ddf85d0a8924 |
line source
3 from allpy.processors import Needle, Left
4 from allpy.fileio import FastaFile
5 from collections import deque
10 def has_identity(column):
11 as_list = column.values()
12 return len(column) == 2 and as_list[0] == as_list[1]
14 def is_good_window(window):
15 sum_id = sum(int(has_identity(column)) for column in window)
16 return len(window) == width and sum_id >= threshold
18 def find_runs(alignment):
19 window = deque([], width)
22 for column in alignment.columns:
24 in_block, was_in_block = is_good_window(window), in_block
25 if in_block and not was_in_block:
26 block = dna.Block.from_alignment(alignment, columns=list(window))
29 block.columns.append(column)
32 def blocks_markup(alignment, blocks):
33 for column in alignment.columns:
36 for column in block.columns:
38 return "".join(column.in_block for column in alignment.columns)
41 alignment = dna.Alignment().append_file(sys.stdin)
42 assert len(alignment.sequences) == 2, "Input must have TWO sequences!"
43 alignment.realign(Left())
44 alignment.realign(Needle())
45 blocks = find_runs(alignment)
47 for n, block in enumerate(blocks, 1):
48 block.to_file(open("block_%02d.fasta" % n, "w"))
50 alignment.to_file(sys.stdout)
51 FastaFile(sys.stdout).write_string(
52 blocks_markup(alignment, blocks),
54 "In run with window %s and threshold %s" % (width, threshold)
60 print "An error has occured:", e
