def build_seqsinfo_from_seqfile(self, seq_file, data_parser_options, num_seqs=numpy.inf):
"""prepares and process sequences to disk and return info dictionary about the parsed sequences
Args:
seq_file: string representing the path to the sequence file
data_parser_options: dictionary containing options to be passed
to :func:`read_file` method of :class:`DataFileParser` class
num_seqs: integer, maximum number of sequences to read from file
(default numpy.inf -- means read all file)
"""
seq_representer = self.seq_representer
# create working directory
corpus_name = "reference_corpus_" + generate_datetime_str()
working_dir = create_directory("working_dir", self.root_dir)
self.working_dir = working_dir
unique_id = False
# build the seqs_info by parsing the sequences from file iteratively
seqs_info = {}
counter=1
dparser = DataFileParser()
for seq in self.get_seqs_from_file(seq_file, dparser, data_parser_options):
if(hasattr(seq, 'id')):
seq_id = seq.id
else:
seq_id = counter
seqs_info.update(seq_representer.prepare_seqs({seq_id:seq}, corpus_name, working_dir, unique_id, log_progress=False))
print("{} sequences have been processed".format(counter))
if(counter>=num_seqs):
break
counter+=1
return(seqs_info)
def build_seqsinfo(self, seq_file):
seq_representer = self.seq_representer
# create working directory
corpus_name = "reference_corpus_" + generate_datetime_str()
working_dir = create_directory("working_dir", self.root_dir)
self.working_dir = working_dir
unique_id = False
# build the seqs_info by parsing the sequences from file iteratively
seqs_info = {}
counter = 1
for seq in self.get_seqs_from_file(seq_file):
if hasattr(seq, "id"):
seq_id = seq.id
else:
seq_id = counter
seqs_info.update(
seq_representer.prepare_seqs(
{seq_id: seq},
corpus_name,
working_dir,
unique_id,
log_progress=False,
)
)
print("{} sequences have been processed".format(counter))
counter += 1
return seqs_info
def use_model(self, savedmodel_dir, options):
"""use trained model for decoding and performance measure evaluation"""
# load learned models
model_dir = savedmodel_dir
# revive/generate learned model
crf_model = self.get_learned_crf(model_dir)
# parse the arguments in kwargs
seqbatch_size = options.get("seqbatch_size")
if(not seqbatch_size):
seqbatch_size = 1000
# check if model evaluation is requested
model_eval = options.get('model_eval')
if(model_eval):
evaluator = SeqDecodingEvaluator(crf_model.model)
perf_metric = options.get('metric')
if(not perf_metric):
perf_metric = 'f1'
exclude_states = options.get('exclude_states')
if(not exclude_states):
exclude_states = []
if(options.get('seqs_info')):
# decode sequences
seqs_info = options.get('seqs_info')
seqs_id = list(seqs_info.keys())
start_ind = 0
stop_ind = seqbatch_size
while(start_ind<len(seqs_id)):
batch_seqsinfo = {seq_id:seqs_info[seq_id] for seq_id in seqs_id[start_ind:stop_ind]}
seqs_pred = crf_model.decode_seqs("viterbi", model_dir, seqs_info=batch_seqsinfo,
file_name=options.get('file_name'), sep=options.get('sep'),
beam_size=options.get('beam_size'))
if(model_eval):
Y_seqs_dict = self.map_pred_to_ref_seqs(seqs_pred)
if(start_ind == 0):
taglevel_perf = evaluator.compute_states_confmatrix(Y_seqs_dict)
else:
taglevel_perf += evaluator.compute_states_confmatrix(Y_seqs_dict)
start_ind+=seqbatch_size
stop_ind+=seqbatch_size
# TO adjust the batch size and available sequences..
elif(options.get('seq_file')):
flag = False
seq_file = options.get('seq_file')
data_parser_options = options.get("data_parser_options")
num_seqs = options.get("num_seqs")
if(not num_seqs):
num_seqs = numpy.inf
# the folder name where intermediary sequences and data are stored
procseqs_foldername = "processed_seqs_" + generate_datetime_str()
dparser = DataFileParser()
seqs_dict = {}
bcounter = 1
seq_counter = 1
for seq in self.get_seqs_from_file(seq_file, dparser, data_parser_options):
seqs_dict[seq_counter] = seq
if(bcounter >= seqbatch_size):
seqs_pred = crf_model.decode_seqs("viterbi", model_dir, seqs_dict=seqs_dict,
procseqs_foldername=procseqs_foldername, file_name=options.get('file_name'),
sep=options.get('sep'), beam_size=options.get('beam_size'))
bcounter = 0
seqs_dict.clear()
if(model_eval):
Y_seqs_dict = self.map_pred_to_ref_seqs(seqs_pred)
if(seq_counter == seqbatch_size):
taglevel_perf = evaluator.compute_states_confmatrix(Y_seqs_dict)
flag = True
else:
taglevel_perf += evaluator.compute_states_confmatrix(Y_seqs_dict)
if(seq_counter>=num_seqs):
break
bcounter += 1
seq_counter+=1
if(len(seqs_dict)):
# decode the remaining sequences
seqs_pred = crf_model.decode_seqs("viterbi", model_dir, seqs_dict=seqs_dict,
procseqs_foldername=procseqs_foldername, file_name=options.get('file_name'),
sep=options.get('sep'), beam_size=options.get('beam_size'))
if(model_eval):
Y_seqs_dict = self.map_pred_to_ref_seqs(seqs_pred)
if(flag):
taglevel_perf += evaluator.compute_states_confmatrix(Y_seqs_dict)
else:
taglevel_perf = evaluator.compute_states_confmatrix(Y_seqs_dict)
if(model_eval):
performance = evaluator.get_performance_metric(taglevel_perf, perf_metric, exclude_states=exclude_states)
return({perf_metric:performance, 'taglevel_confusion_matrix':taglevel_perf})
return({})
def eval_model(self, savedmodel_dir, options):
# load learned models
model_dir = savedmodel_dir
# revive/generate learned model
crf_model = self.get_learned_crf(model_dir)
# parse the arguments in kwargs
seqbatch_size = options.get("seqbatch_size")
if not seqbatch_size:
seqbatch_size = 1000
# check if model evaluation is requested
model_eval = options.get("model_eval")
if model_eval:
evaluator = SeqDecodingEvaluator(crf_model.model)
perf_metric = options.get("metric")
if not perf_metric:
perf_metric = "f1"
exclude_states = options.get("exclude_states")
if not exclude_states:
exclude_states = []
if options.get("seqs_info"):
# decode sequences
seqs_info = options.get("seqs_info")
seqs_id = list(seqs_info.keys())
start_ind = 0
stop_ind = seqbatch_size
while start_ind < len(seqs_id):
batch_seqsinfo = {
seq_id: seqs_info[seq_id] for seq_id in seqs_id[start_ind:stop_ind]
}
seqs_pred = crf_model.decode_seqs(
"viterbi",
model_dir,
seqs_info=batch_seqsinfo,
file_name=options.get("file_name"),
sep=options.get("sep"),
beam_size=options.get("beam_size"),
)
if model_eval:
Y_seqs_dict = self.map_pred_to_ref_seqs(seqs_pred)
if start_ind == 0:
taglevel_perf = evaluator.compute_states_confmatrix(Y_seqs_dict)
else:
taglevel_perf += evaluator.compute_states_confmatrix(
Y_seqs_dict
)
start_ind += seqbatch_size
stop_ind += seqbatch_size
# TO adjust the batch size and available sequences..
elif options.get("seq_file"):
flag = False
seq_file = options.get("seq_file")
# the folder name where intermediary sequences and data are stored
procseqs_foldername = "processed_seqs_" + generate_datetime_str()
seqs_dict = {}
bcounter = 1
seq_counter = 1
for seq in self.get_seqs_from_file(seq_file):
seqs_dict[seq_counter] = seq
if bcounter >= seqbatch_size:
seqs_pred = crf_model.decode_seqs(
"viterbi",
model_dir,
seqs_dict=seqs_dict,
procseqs_foldername=procseqs_foldername,
file_name=options.get("file_name"),
sep=options.get("sep"),
beam_size=options.get("beam_size"),
)
bcounter = 0
seqs_dict.clear()
if model_eval:
Y_seqs_dict = self.map_pred_to_ref_seqs(seqs_pred)
if seq_counter == seqbatch_size:
taglevel_perf = evaluator.compute_states_confmatrix(
Y_seqs_dict
)
flag = True
else:
taglevel_perf += evaluator.compute_states_confmatrix(
Y_seqs_dict
)
bcounter += 1
seq_counter += 1
if len(seqs_dict):
# decode the remaining sequences
seqs_pred = crf_model.decode_seqs(
"viterbi",
model_dir,
seqs_dict=seqs_dict,
procseqs_foldername=procseqs_foldername,
file_name=options.get("file_name"),
sep=options.get("sep"),
beam_size=options.get("beam_size"),
)
if model_eval:
Y_seqs_dict = self.map_pred_to_ref_seqs(seqs_pred)
if flag:
taglevel_perf += evaluator.compute_states_confmatrix(
Y_seqs_dict
)
else:
taglevel_perf = evaluator.compute_states_confmatrix(Y_seqs_dict)
if model_eval:
performance = evaluator.get_performance_metric(
taglevel_perf, perf_metric, exclude_states=exclude_states
)
return {
perf_metric: performance,
"taglevel_confusion_matrix": taglevel_perf,
}
return {}