allpy: fdc2d3594a3e test/usecase1.py

allpy

view test/usecase1.py @ 1080:fdc2d3594a3e

allpy.base: Added decency verification function to alignments (closes #98) Currently, the verification function asserts that no two sequences in an alignment have the same identifier.

author	Daniil Alexeyevsky <dendik@kodomo.fbb.msu.ru>
date	Wed, 30 May 2012 12:43:09 +0400
parents	af1de9132035
children	08d892230e8c

line source

1 #!/usr/bin/python

2 import sys

3 from allpy import protein

4 from allpy import processors

6 # Create sequences from string representation of sequence body

7 sequence_1 = protein.Sequence.from_string("mkstf", name="E2E4")

8 sequence_2 = protein.Sequence.from_string("mstkfff", description="Longer sequence")

10 # Create alignment from sequences

11 alignment = protein.Alignment()

12 alignment.append_sequence(sequence_1)

13 alignment.append_sequence(sequence_2)

14 alignment.realign(processors.Muscle())

16 # For each sequence, print number of gaps and non-gaps in alignment

17 for row in alignment.rows():

18 gaps = 0

19 monomers = 0

20 for column in alignment.columns:

21 if column in row:

22 monomers += 1

23 else:

24 gaps += 1

25 print "%s: %s gaps, %s non-gaps" % (row.sequence.name, gaps, monomers)

27 # Print number of gaps in each column

28 gaps = []

29 for column in alignment.columns:

30 column_gaps = 0

31 for sequence in alignment.sequences:

32 if sequence not in column:

33 column_gaps += 1

34 gaps.append(column_gaps)

35 print " ".join(map(str, gaps))

37 # Write alignment to file

38 alignment.to_file(sys.stdout)