allpy: 936463fb68b2 allpy/fileio.py

allpy

view allpy/fileio.py @ 506:936463fb68b2

fixed blocks3d() * fix bug in continuous_blocks it joined short batches to next batch * blocks3d(): put monomers only of boundary columns to the graph. * reduce size of wide cliques * optimization this needs better solution * corrected conversion clique -> blocks (result) * fixed self-intersection of result blocks (filled_columns) * fixed infinite loop * added temponary exclusion of shortest sequence * sorting of cliques by height and by width

author	boris (kodomo) <bnagaev@gmail.com>
date	Wed, 23 Feb 2011 18:13:56 +0300
parents	0bd118c2d72b
children	5dfb9b9761d5

line source

1 import os

2 from tempfile import NamedTemporaryFile

4 import util

6 class BaseIo(object):

7 """ Base class providing alignment/sequence import and export

9 Data:

10 * file - file object

11 """

13 def __init__(self, file):

14 self.file = file

16 def save_string(self, string, name, description=''):

17 """ Saves given string to file

19 Splits long lines to substrings of length=long_line

20 To prevent this, set long_line=None

21 """

22 pass

24 def get_all_strings(self):

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

27 Return a list of tuples (name, description, sequence_body).

28 """

29 pass

31 def get_string(self, name):

32 """ return tuple (name, description, string) for sequence with name name """

33 for name_test, description, body in self.get_all_strings():

34 if name_test == name:

35 return (name_test, description, body)

37 class FastaIo(BaseIo):

38 """ Fasta import and export

40 Additional data:

41 * long_line - max length of file line while export

42 Splits long lines to substrings of length=long_line

43 To prevent this, set long_line=None

44 """

46 def __init__(self, file, long_line=70):

47 BaseIo.__init__(self, file)

48 self.long_line = long_line

50 def save_string(self, string, name, description=''):

51 if description:

52 name += " " + description

53 self.file.write(">%s\n" % name)

54 if self.long_line:

55 for i in range(0, len(string) // self.long_line + 1):

56 start = i*self.long_line

57 end = i*self.long_line + self.long_line

58 self.file.write("%s\n" % string[start:end])

59 else:

60 self.file.write("%s\n" % string)

62 def get_all_strings(self):

63 for part in self.file.read().split("\n>"):

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

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

66 name, _, description = header.partition(" ")

67 name = name.strip()

68 description = description.strip()

69 body = util.remove_each(body, " \n\r\t\v")

70 yield (name, description, body)

72 def get_string(self, name):

73 for name_test, description, body in self.get_all_strings():

74 if name_test == name:

75 return (name_test, description, body)

77 class MsfIo(BaseIo):

78 """ Msf import and export """

80 def __init__(self, file):

81 BaseIo.__init__(self, file)

83 def save_string(self, string, name, description=''):

84 tmp_fasta = NamedTemporaryFile('w', delete=False)

85 fasta = FastaIo(tmp_fasta)

86 fasta.save_string(string, name, description)

87 tmp_fasta.close()

88 self.file.close()

89 os.system("seqret %(fasta)s msf::%(msf)s" % \

90 {'msf': self.file.name, 'fasta': tmp_fasta.name})

91 os.unlink(tmp_fasta.name)

92 self.file = open(self.file.name)

94 def get_all_strings(self):

95 tmp_fasta = NamedTemporaryFile(delete=False)

96 os.system("seqret %(msf)s %(fasta)s" % \

97 {'msf': self.file.name, 'fasta': tmp_fasta.name})

98 fasta = FastaIo(tmp_fasta)

99 strings = list(fasta.get_all_strings())

100 os.unlink(tmp_fasta.name)

101 return strings

102