def parse(handle, format):
"""Parses an output file of motif finding programs.
Currently supported formats:
- AlignAce
- MEME
You can also use single-motif formats, although the Bio.Motif.read()
function is simpler to use in this situation.
- jaspar-pfm
- jaspar-sites
For example:
>>> from Bio import Motif
>>> for motif in Motif.parse(open("Motif/alignace.out"),"AlignAce"):
... print motif.consensus()
TCTACGATTGAG
CTGCACCTAGCTACGAGTGAG
GTGCCCTAAGCATACTAGGCG
GCCACTAGCAGAGCAGGGGGC
CGACTCAGAGGTT
CCACGCTAAGAGAAGTGCCGGAG
GCACGTCCCTGAGCA
GTCCATCGCAAAGCGTGGGGC
GAGATCAGAGGGCCG
TGGACGCGGGG
GACCAGAGCCTCGCATGGGGG
AGCGCGCGTG
GCCGGTTGCTGTTCATTAGG
ACCGACGGCAGCTAAAAGGG
GACGCCGGGGAT
CGACTCGCGCTTACAAGG
"""
if format in ('pfm', 'sites'):
yield Jaspar.read(handle, format)
else:
try:
parser = _parsers[format]
except KeyError:
try: #not a true parser, try reader formats
reader = _readers[format]
except:
raise ValueError("Wrong parser format")
else: #we have a proper reader
yield reader(handle)
else: # we have a proper reader
for m in parser(handle).motifs:
yield m
def parse(handle, format):
"""Parses an output file of motif finding programs.
Currently supported formats:
- AlignAce
- MEME
You can also use single-motif formats, although the Bio.Motif.read()
function is simpler to use in this situation.
- jaspar-pfm
- jaspar-sites
For example:
>>> from Bio import Motif
>>> for motif in Motif.parse(open("Motif/alignace.out"),"AlignAce"):
... print motif.consensus
TCTACGATTGAG
CTGCAGCTAGCTACGAGTGAG
GTGCTCTAAGCATAGTAGGCG
GCCACTAGCAGAGCAGGGGGC
CGACTCAGAGGTT
CCACGCTAAGAGAGGTGCCGGAG
GCGCGTCGCTGAGCA
GTCCATCGCAAAGCGTGGGGC
GGGATCAGAGGGCCG
TGGAGGCGGGG
GACCAGAGCTTCGCATGGGGG
GGCGTGCGTG
GCTGGTTGCTGTTCATTAGG
GCCGGCGGCAGCTAAAAGGG
GAGGCCGGGGAT
CGACTCGTGCTTAGAAGG
"""
if format in ("pfm", "sites"):
yield Jaspar.read(handle, format)
else:
try:
parser = _parsers[format]
except KeyError:
try: # not a true parser, try reader formats
reader = _readers[format]
except:
raise ValueError("Wrong parser format")
else: # we have a proper reader
yield reader(handle)
else: # we have a proper reader
for m in parser(handle).motifs:
yield m
def format(self,format):
"""Returns a string representation of the Motif in a given format
Currently supported fromats:
- pfm : JASPAR Position Frequency Matrix
- transfac : TRANSFAC like files
"""
if format=="pfm":
from Bio.Motif import Jaspar
return Jaspar.write(self)
elif format=="transfac":
from Bio.Motif import TRANSFAC
motifs = [self]
return TRANSFAC.write(motifs)
else:
raise ValueError("Unknown format type %s" % format)
def write(motifs, format):
"""Returns a string representation of motifs in a given format
Currently supported fromats:
- pfm : JASPAR Position Frequency Matrix
[only if len(motifs)==1]
- transfac : TRANSFAC like files
"""
if format=="pfm":
from Bio.Motif import Jaspar
if len(motifs)!=1:
raise Exception("Only a single motif can be written in the JASPAR Position Frequency Matrix (pfm) format")
motif = motifs[0]
return Jaspar.write(motif)
elif format=="transfac":
from Bio.Motif import TRANSFAC
return TRANSFAC.write(motifs)
else:
raise ValueError("Unknown format type %s" % format)
def parse(handle, format):
"""Parses an output file of motif finding programs.
Currently supported formats:
- AlignAce: AlignAce output file format
- MEME: MEME output file motif
- TRANSFAC: TRANSFAC database file format
- pfm: JASPAR-style position-frequency matrix
- sites: JASPAR-style sites file
- jaspar-pfm: JASPAR-style position-frequency matrix [DEPRECATED]
- jaspar-sites: JASPAR-style sites file [DEPRECATED]
As files in the pfm and sites formats contain only a single motif,
it is easier to use Bio.Motif.read() instead of Bio.Motif.parse()
for those.
For example:
>>> from Bio import Motif
>>> for motif in Motif.parse(open("Motif/alignace.out"),"AlignAce"):
... print motif.consensus()
TCTACGATTGAG
CTGCACCTAGCTACGAGTGAG
GTGCCCTAAGCATACTAGGCG
GCCACTAGCAGAGCAGGGGGC
CGACTCAGAGGTT
CCACGCTAAGAGAAGTGCCGGAG
GCACGTCCCTGAGCA
GTCCATCGCAAAGCGTGGGGC
GAGATCAGAGGGCCG
TGGACGCGGGG
GACCAGAGCCTCGCATGGGGG
AGCGCGCGTG
GCCGGTTGCTGTTCATTAGG
ACCGACGGCAGCTAAAAGGG
GACGCCGGGGAT
CGACTCGCGCTTACAAGG
>>> for motif in Motif.parse(open("Motif/alignace.out"),"alignace"):
... print motif.consensus
TCTACGATTGAG
CTGCAGCTAGCTACGAGTGAG
GTGCTCTAAGCATAGTAGGCG
GCCACTAGCAGAGCAGGGGGC
CGACTCAGAGGTT
CCACGCTAAGAGAGGTGCCGGAG
GCGCGTCGCTGAGCA
GTCCATCGCAAAGCGTGGGGC
GGGATCAGAGGGCCG
TGGAGGCGGGG
GACCAGAGCTTCGCATGGGGG
GGCGTGCGTG
GCTGGTTGCTGTTCATTAGG
GCCGGCGGCAGCTAAAAGGG
GAGGCCGGGGAT
CGACTCGTGCTTAGAAGG
"""
if format=="AlignAce":
# Old Motif code
from Bio.Motif.Parsers import AlignAce
record = AlignAce.read(handle)
return iter(record.motifs)
elif format=="alignace":
# Old Motif code
from Bio.Motif import AlignAce
record = AlignAce.read(handle)
return record
elif format=="MEME":
from Bio.Motif.Parsers import MEME
record = MEME.read(handle)
return iter(record.motifs)
elif format=="meme":
from Bio.Motif import MEME
record = MEME.read(handle)
return record
elif format=="TRANSFAC":
from Bio.Motif import TRANSFAC
record = TRANSFAC.read(handle)
return record
elif format in ('pfm', 'sites'):
from Bio.Motif import Jaspar
motif = Jaspar.read(handle, format)
elif format=="jaspar-pfm":
motif = OldMotif()._from_jaspar_pfm(handle)
return iter([motif])
elif format=="jaspar-sites":
motif = OldMotif()._from_jaspar_sites(handle)
return iter([motif])
else:
raise ValueError("Unknown format %s" % format)
# Treat the single-motif formats
motifs = [motif]
return motifs
