allpy: 2ede0318ff95 test/usecase3.py

allpy

Added virtual monomer types for undefined letters N and X, changed all previous such monomers to be considered modified

author	Daniil Alexeyevsky <me.dendik@gmail.com>
date	Fri, 21 Jan 2011 17:01:21 +0300
parents	20944c113ea3
children	387f692e2356

line source

1 from allpy import protein

2 alignment = protein.Alignment.from_file(open("aln.fasta"))

3 #conservative = [(10,20), (40,50)]

4 conservative = [(0,6),(18,37)]

6 def ranges_to_markup(ranges):

7 """Convert list of ranges to line of markup.

9 This has nothing to do with allpy.

10 """

11 markup = ["-"] * len(alignment.columns)

12 for begin, end in ranges:

13 for i in range(begin, end+1):

14 markup[i] = "+"

15 return "".join(markup)

17 def markup_to_blocks(markup):

18 """Convert markup line to a bunch of blocks, one for each sequential run."""

19 current = None

20 blocks = {}

21 for mark, column in zip(markup, alignment.columns):

22 if mark != current:

23 block = protein.Block.from_alignment(alignment, columns=[])

24 blocks[mark] = blocks.get(mark, []) + [block]

25 current = mark

26 blocks[mark][-1].columns.append(column)

27 return blocks

29 def main():

30 markup = ranges_to_markup(conservative)

31 blocks = markup_to_blocks(markup)

32 for block in blocks["-"]:

33 block.flush_left()

34 alignment.to_fasta(open("output.fasta", "w"))

36 main()