class MultinomialModelRunner(ModelRunner):
def run(self, args):
self.reader = MultinomialCountsReader(args.mcnt_file_name)
self.writer = JointMultiMixWriter(args.jmm_file_name)
ModelRunner.run(self, args)
def _train(self, args):
if args.subsample_size > 0:
counts = self._subsample(args.subsample_size)
else:
counts = self.reader.get_counts()
self.priors_parser.load_from_file(args.priors_file)
self.priors = self.priors_parser.to_dict()
self._write_priors()
data = MultinomialData(counts)
self.parameters = self.model.train(data, self.priors,
args.max_iters, args.convergence_threshold)
def _classify_chromosome(self, chr_name):
counts = self.reader.get_counts(chr_name)
jcnt_rows = self.reader.get_rows(chr_name)
end = self.reader.get_chr_size(chr_name)
n = int(1e5)
start = 0
stop = min(n, end)
while start < end:
sub_counts = counts[start:stop]
sub_rows = jcnt_rows[start:stop]
data = MultinomialData(sub_counts)
resp = self.model.classify(data, self.parameters)
self.writer.write_data(chr_name, sub_rows, resp)
start = stop
stop = min(stop + n, end)
def run(self, args):
self.reader = MultinomialCountsReader(args.mcnt_file_name)
self.writer = JointMultiMixWriter(args.jmm_file_name)
ModelRunner.run(self, args)