def cache_parameters(self):
assert self.args.n_sets == 1 # er, could do it for n > 1, but I'd want to think through a few things first
assert self.args.plotdir is not None
sw_parameter_dir = self.args.parameter_dir + '/sw'
waterer = Waterer(self.args,
self.input_info,
self.reco_info,
self.germline_seqs,
parameter_dir=sw_parameter_dir,
write_parameters=True,
plotdir=self.args.plotdir + '/sw')
waterer.run()
self.write_hmms(sw_parameter_dir, waterer.info['all_best_matches'])
parameter_in_dir = sw_parameter_dir
for ibw in range(self.args.baum_welch_iterations):
parameter_out_dir = self.args.parameter_dir + '/hmm'
hmm_plotdir = self.args.plotdir + '/hmm'
if self.args.baum_welch_iterations > 1:
parameter_out_dir += '-' + str(ibw)
hmm_plotdir += '-' + str(ibw)
self.run_hmm('viterbi',
waterer.info,
parameter_in_dir=parameter_in_dir,
parameter_out_dir=parameter_out_dir,
hmm_type='k=1',
count_parameters=True,
plotdir=hmm_plotdir)
self.write_hmms(parameter_out_dir,
waterer.info['all_best_matches'])
parameter_in_dir = parameter_out_dir
if not self.args.no_clean:
os.rmdir(self.args.workdir)
def run_algorithm(self, algorithm):
if not os.path.exists(self.args.parameter_dir):
raise Exception('ERROR ' + self.args.parameter_dir + ' d.n.e')
waterer = Waterer(self.args, self.input_info, self.reco_info, self.germline_seqs, parameter_dir=self.args.parameter_dir, write_parameters=False)
waterer.run()
self.run_hmm(algorithm, waterer.info, parameter_in_dir=self.args.parameter_dir, hmm_type='k=nsets', \
count_parameters=self.args.plot_parameters, plotdir=self.args.plotdir)
# self.clean(waterer)
if not self.args.no_clean:
os.rmdir(self.args.workdir)
def run_algorithm(self, algorithm):
if not os.path.exists(self.args.parameter_dir):
raise Exception('ERROR ' + self.args.parameter_dir + ' d.n.e')
waterer = Waterer(self.args,
self.input_info,
self.reco_info,
self.germline_seqs,
parameter_dir=self.args.parameter_dir,
write_parameters=False)
waterer.run()
self.run_hmm(algorithm, waterer.info, parameter_in_dir=self.args.parameter_dir, hmm_type='k=nsets', \
count_parameters=self.args.plot_parameters, plotdir=self.args.plotdir)
# self.clean(waterer)
if not self.args.no_clean:
os.rmdir(self.args.workdir)
def partition(self):
assert os.path.exists(self.args.parameter_dir)
# run smith-waterman
waterer = Waterer(self.args,
self.input_info,
self.reco_info,
self.germline_seqs,
parameter_dir=self.args.parameter_dir,
write_parameters=False)
waterer.run()
# cdr3 length partitioning
cdr3_cluster = False # don't precluster on cdr3 length for the moment -- I cannot accurately infer cdr3 length in some sequences, so I need a way to pass query seqs to the clusterer with several possible cdr3 lengths (and I don't know how to do that!)
cdr3_length_clusters = None
if cdr3_cluster:
cdr3_length_clusters = self.cdr3_length_precluster(waterer)
hamming_clusters = self.hamming_precluster(cdr3_length_clusters)
# stripped_clusters = self.run_hmm('forward', waterer.info, self.args.parameter_dir, preclusters=hamming_clusters, stripped=True)
hmm_clusters = self.run_hmm('forward',
waterer.info,
self.args.parameter_dir,
preclusters=hamming_clusters,
hmm_type='k=2',
make_clusters=True)
self.run_hmm('forward',
waterer.info,
self.args.parameter_dir,
preclusters=hmm_clusters,
hmm_type='k=preclusters',
prefix='k-',
make_clusters=False)
# self.run_hmm('viterbi', waterer.info, self.args.parameter_dir, preclusters=hmm_clusters, hmm_type='k=preclusters', prefix='k-', make_clusters=False)
# self.clean(waterer)
if not self.args.no_clean:
os.rmdir(self.args.workdir)
def cache_parameters(self):
assert self.args.n_sets == 1 # er, could do it for n > 1, but I'd want to think through a few things first
assert self.args.plotdir is not None
sw_parameter_dir = self.args.parameter_dir + '/sw'
waterer = Waterer(self.args, self.input_info, self.reco_info, self.germline_seqs, parameter_dir=sw_parameter_dir, write_parameters=True, plotdir=self.args.plotdir + '/sw')
waterer.run()
self.write_hmms(sw_parameter_dir, waterer.info['all_best_matches'])
parameter_in_dir = sw_parameter_dir
for ibw in range(self.args.baum_welch_iterations):
parameter_out_dir = self.args.parameter_dir + '/hmm'
hmm_plotdir = self.args.plotdir + '/hmm'
if self.args.baum_welch_iterations > 1:
parameter_out_dir += '-' + str(ibw)
hmm_plotdir += '-' + str(ibw)
self.run_hmm('viterbi', waterer.info, parameter_in_dir=parameter_in_dir, parameter_out_dir=parameter_out_dir, hmm_type='k=1', count_parameters=True, plotdir=hmm_plotdir)
self.write_hmms(parameter_out_dir, waterer.info['all_best_matches'])
parameter_in_dir = parameter_out_dir
if not self.args.no_clean:
os.rmdir(self.args.workdir)
def partition(self):
assert os.path.exists(self.args.parameter_dir)
# run smith-waterman
waterer = Waterer(self.args, self.input_info, self.reco_info, self.germline_seqs, parameter_dir=self.args.parameter_dir, write_parameters=False)
waterer.run()
# cdr3 length partitioning
cdr3_cluster = False # don't precluster on cdr3 length for the moment -- I cannot accurately infer cdr3 length in some sequences, so I need a way to pass query seqs to the clusterer with several possible cdr3 lengths (and I don't know how to do that!)
cdr3_length_clusters = None
if cdr3_cluster:
cdr3_length_clusters = self.cdr3_length_precluster(waterer)
hamming_clusters = self.hamming_precluster(cdr3_length_clusters)
# stripped_clusters = self.run_hmm('forward', waterer.info, self.args.parameter_dir, preclusters=hamming_clusters, stripped=True)
hmm_clusters = self.run_hmm('forward', waterer.info, self.args.parameter_dir, preclusters=hamming_clusters, hmm_type='k=2', make_clusters=True)
self.run_hmm('forward', waterer.info, self.args.parameter_dir, preclusters=hmm_clusters, hmm_type='k=preclusters', prefix='k-', make_clusters=False)
# self.run_hmm('viterbi', waterer.info, self.args.parameter_dir, preclusters=hmm_clusters, hmm_type='k=preclusters', prefix='k-', make_clusters=False)
# self.clean(waterer)
if not self.args.no_clean:
os.rmdir(self.args.workdir)
