def testWriteStockholm(self):
stock = StringIO()
with MSAFile(stock, 'w', format='stockholm') as out:
for seq in MSAFile(pathDatafile('msa_Cys_knot.sth'), split=False):
out.write(seq)
stock.seek(0)
stock_list = list(MSAFile(stock, format='stockholm'))
self.assertListEqual(STOCK_LIST, stock_list)
def testWriteSelex(self):
selex = StringIO()
with MSAFile(selex, 'w', format='selex') as out:
for seq in MSAFile(pathDatafile('msa_Cys_knot.slx')):
out.write(seq)
selex.seek(0)
selex_list = list(MSAFile(selex, format='selex'))
self.assertListEqual(SELEX_LIST, selex_list)
def testWriteFasta(self):
fasta = StringIO()
with MSAFile(fasta, 'w', format='fasta') as out:
for seq in MSAFile(pathDatafile('msa_Cys_knot.fasta')):
out.write(seq)
fasta.seek(0)
fasta_list = list(MSAFile(fasta, format='fasta'))
self.assertListEqual(FASTA_LIST, fasta_list)
def evol_merge(*msa, **kwargs):
import prody
from prody import parseMSA, mergeMSA, LOGGER, writeMSA, MSAFile
from prody.sequence.msafile import MSAEXTMAP
from os.path import splitext
if len(msa) < 2:
raise ValueError('multiple msa filenames must be specified')
msaobj = []
try:
msaobj = [parseMSA(fn) for fn in msa]
except:
raise IOError('failed to parse {0}'.format(fn))
msafile = MSAFile(msa[0])
format = kwargs.get('format') or msafile.format
outname = kwargs.get('outname') or (msafile.getTitle() + '_merged' +
MSAEXTMAP[msafile.format])
writeMSA(outname, mergeMSA(*msaobj), **kwargs)
LOGGER.info('Merged MSA is saved as: {0}'.format(outname))
def evol_filter(msa, *word, **kwargs):
import prody
from prody import MSAFile, writeMSA, LOGGER
from os.path import splitext
outname = kwargs.get('outname')
if outname is None:
outname, ext = splitext(msa)
if ext.lower() == '.gz':
outname, _ = splitext(msa)
outname += '_filtered' + ext
single = len(word) == 1
if single:
word = word[0]
if kwargs.get('startswith', False):
if single:
filter = lambda label, seq, word=word: label.startswith(word)
elif kwargs.get('endswith', False):
if single:
filter = lambda label, seq, word=word: label.endswith(word)
elif kwargs.get('contains', False):
if single:
filter = lambda label, seq, word=word: word in label
elif kwargs.get('equals', False):
if single:
filter = lambda label, seq, word=word: word == label
else:
filter = lambda label, seq, word=set(word): label in word
else:
raise TypeError('one of startswith, endswith, contains, or equals '
'must be specified')
msa = MSAFile(msa,
filter=filter,
filter_full=kwargs.get('filter_full', False))
LOGGER.info('Filtered MSA is written in file: ' +
writeMSA(outname, msa, **kwargs))
def testFasta(self):
filename = writeMSA(join(TEMPDIR, 'test.fasta.gz'), FASTA)
fasta = list(MSAFile(pathDatafile(filename)))
self.assertListEqual(list(FASTA), list(fasta))
if os.path.isfile(filename):
os.remove(filename)
except ImportError:
from io import StringIO
from numpy import array, log, zeros, char
from numpy.testing import assert_array_equal, dec
from prody.tests.datafiles import *
from prody.tests import TEMPDIR
from prody import MSA, MSAFile, parseMSA, LOGGER, writeMSA
LOGGER.verbosity = None
FASTA = parseMSA(pathDatafile('msa_Cys_knot.fasta'))
SELEX = parseMSA(pathDatafile('msa_Cys_knot.slx'))
STOCK = parseMSA(pathDatafile('msa_Cys_knot.sth'))
FASTA_LIST = list(MSAFile(pathDatafile('msa_Cys_knot.fasta')))
SELEX_LIST = list(MSAFile(pathDatafile('msa_Cys_knot.sth')))
STOCK_LIST = list(MSAFile(pathDatafile('msa_Cys_knot.sth')))
class TestMSAFile(TestCase):
def testMSAFile(self):
self.assertListEqual(FASTA_LIST, SELEX_LIST)
self.assertListEqual(FASTA_LIST, STOCK_LIST)
def testWriteFasta(self):
fasta = StringIO()
with MSAFile(fasta, 'w', format='fasta') as out:
for seq in MSAFile(pathDatafile('msa_Cys_knot.fasta')):