allpy: 2b3cad50c2b1 utils/freqs.py

allpy

view utils/freqs.py @ 1106:2b3cad50c2b1

Partially reversed [afed1f] (see #49) As explained in the ticket, in real life usecases having a monomer belong to several sequences is sometimes extremely useful. ANY approach to attribution of monomer to only one sequence will be either confusing or hindering. * Removed `monomer.sequence` attribute * Removed unncecessary specialcasing in pickle * Removed unused tests * Restored APIs to backward-compatible * Added deprecated messages to the restored APIs

author	Daniil Alexeyevsky <dendik@kodomo.fbb.msu.ru>
date	Sun, 10 Jun 2012 16:08:47 +0400
parents	8c7e5c16b2f4
children

line source

1 #!/usr/bin/python

2 """Read alignment on stdin. Print CSV table of letter frequences on stdout.

3 """

4 from allpy import protein

5 from allpy.data import codes

6 import sys

8 sys.stderr.write(__doc__)

10 def freq(monomer):

11 amount = freqs.get(monomer)

12 if amount:

13 return 100.0 * amount / width

14 return ""

16 aln = protein.Alignment().append_file(sys.stdin)

17 monomers = [code1 for code1, modified, _, _ in codes.protein if not modified]

18 monomers += ["-"]

19 width = len(aln.sequences)

20 print ", ".join(map(str, monomers))

21 for column in aln.columns_as_lists():

22 freqs = {}

23 for monomer in column:

24 if monomer:

25 monomer = monomer.code1

26 else:

27 monomer = "-"

28 freqs[monomer] = freqs.get(monomer, 0) + 1

29 print ", ".join(map(str, map(freq, monomers)))