allpy: cf6cdc3b7ec5 allpy/fasta.py

allpy

Fixed spaces

author	Daniil Alexeyevsky <me.dendik@gmail.com>
date	Wed, 15 Dec 2010 23:45:11 +0300
parents	4e6e85851133
children	f8bd7c469fcf

line source

1 def parse_fasta(file):

2 """Parse fasta file, remove spaces and newlines from sequence bodies.

4 Return a dict of { sequence header: sequence body }.

5 """

6 sequences = {}

7 for part in file.read().split(">"):

8 header, _, body = part.partition("\n")

9 header = header.lstrip(">").strip()

10 body = body.replace(" ", "").replace("\n", "")

11 sequences[header] = body

12 return sequences

14 def save_fasta(out_file, string, name, description='', long_line=70):

15 """ Saves given string to out_file in fasta_format

17 Splits long lines to substrings of length=long_line

18 To prevent this, set long_line=None

19 """

20 out_file.write(">%(name)s %(description)s \n" % {'name':name, 'description':description})

21 if long_line:

22 for i in range(0, len(string) // long_line + 1):

23 out_file.write("%s \n" % string[i*long_line : i*long_line + long_line])

24 else:

25 out_file.write("%s \n" % string)

27 def determine_long_line(in_file):

28 """ Returns maximum sequence line length in fasta file """

29 sequences = in_file.read().split('>')

30 for sequence in sequences[1:]:

31 lines = sequence.split('\n')[1:]

32 if len(lines) >= 2:

33 return len(lines[0].strip())

34 return 70

36 # vim: set ts=4 sts=4 sw=4 et: