def make_master_fasta(self, output_fasta=None, force=None):
'''
Directly create a fasta file from the master sequences of each
cluster of a list.
@param: output_fasta
@pdef: name of the Fasta file
@pdefault: input fasta + threshold + '.fa'
@ptype: {String}
@param: force
@pdef: overwrite previous files with the same name
@pdefault: _SBIglobals.overwrite_
@ptype: {Boolean}
@return: {Fasta}
'''
if output_fasta is None:
output_fasta = self.master_fasta_file
clist = self.parse()
prots = set()
fs = Fasta(self._input)
for c in clist.clusters:
prots.add(c.master.name)
seq = fs.retrieve(prots)
return Fasta.build_multifasta(output_fasta, seq, force=force)
def get_PDBseq_filtered(self, resolution_threshold, output_file):
resolutions = self.get_resolutions()
names = [
k for k, v in resolutions.iteritems()
if float(v) <= float(resolution_threshold)
]
sequences = Fasta(os.path.join(self.PDBseq, 'PDBseq.fa'))
selectedseq = sequences.retrieve(copy.deepcopy(names), prefix_size=4)
return Fasta.build_multifasta(output_file, selectedseq, True)
def get_FASTA_IDX_by_names_to_file(self, names, outfile):
fastafile = Fasta(self.PDBseq)
selectedfasta = fastafile.retrieve(copy.deepcopy(names))
output_fasta = File(outfile, 'w')
for sequence in selectedfasta:
output_fasta.write(sequence.format('FASTA') + "\n")
output_fasta.close()
idxfile = self.PDBseq + '.idx'
output_idx = File(outfile + '.idx', 'w')
input_idx = File(idxfile, 'r')
for line in input_idx.descriptor:
info = line.split()
pdbname = info[0][1:]
if pdbname in names:
output_idx.write(line)
input_idx.close()
output_idx.close()