def _parse_joint_site_frequencies(self, filepath, field_name_prefix,
is_normalize_by_site_counts,
lineage_pair, locus_definition,
results_d):
with utility.universal_open(filepath) as src:
lines = src.read().split("\n")
col_keys = [
c for c in lines[1].split("\t")[1:]
if "sites with multiple" not in c
]
row_idx = 0
for line in lines[2:]:
if not line:
continue
cols = line.split("\t")
if len(cols) - 1 != len(col_keys):
raise ValueError(
"Row {}: Expecting {} columns, but found {}: {}".
format(row_idx + 1, len(col_keys),
len(cols) - 1, cols))
assert len(cols) - 1 == len(col_keys)
row_key = cols[0]
col_idx = 0
for col_key, val in zip(col_keys, cols[1:]):
# results_d["{}.{}.{}".format(field_name_prefix, row_key, col_key)] = float(val)
val = float(val)
if is_normalize_by_site_counts:
val = val / locus_definition.num_sites
results_d["{}.{}.{}".format(field_name_prefix, row_idx,
col_idx)] = val
col_idx += 1
row_idx += 1
return results_d
def _generate_parameter_file(
self,
fsc2_config_d,
):
assert self.parameter_filepath
with utility.universal_open(
os.path.join(self.working_directory, self.parameter_filepath),
"w") as dest:
self._write_parameter_configuration(
dest=dest,
fsc2_config_d=fsc2_config_d,
)
def _parse_raw_results_dna(self):
data_dict = collections.OrderedDict()
decodings = list(itertools.permutations("ACGT", 4))
column_decodings = {}
with utility.universal_open(self.arlequin_filepath) as src:
idx = 0
first_pop_max_ind_id = 0
in_alignment = False
for row in src:
row = row[:-1] # chomp terminating \n
if in_alignment:
if row == "":
if idx == 2:
break
else:
in_alignment = False
else:
try:
x1, x2, encoded_data = row.split("\t")
decoded_data = []
for col_idx, ch in enumerate(encoded_data):
try:
decoding_lookup = column_decodings[col_idx]
except KeyError:
decoding_lookup = self.rng.choice(
decodings)
column_decodings[col_idx] = decoding_lookup
decoded_data.append(decoding_lookup[int(ch)])
pop_id, ind_id = [
int(label_part) for label_part in x1.split("_")
]
# this ugly hack is because fastsimcoal inconsistently numbers
# individuals on trees vs sequences ... even
# assuming there is a correspondence
if pop_id == 1:
first_pop_max_ind_id = max(
first_pop_max_ind_id, ind_id)
else:
ind_id += first_pop_max_ind_id
taxon_label = self._compose_raw_data_taxon_label(
population_id=pop_id, individual_id=ind_id)
# taxon_label = self._compose_raw_data_taxon_label(
# population_id=idx,
# individual_id=x1)
data_dict[taxon_label] = "".join(decoded_data)
except IndexError:
raise
elif "SampleData=" in row:
idx += 1
in_alignment = True
return dendropy.DnaCharacterMatrix.from_dict(data_dict)
def process(self, target_data_filepath, priors_data_filepaths,
output_name_prefix, output_directory, output_suffix):
if output_name_prefix is None:
output_name_prefix = os.path.splitext(
os.path.basename(target_data_filepath))[0]
if output_suffix is None:
output_suffix = ""
else:
output_suffix = "." + output_suffix
with utility.universal_open(target_data_filepath) as src:
target_data_reader = csv.DictReader(src,
delimiter=self.field_delimiter,
quoting=csv.QUOTE_NONE)
for target_row_idx, target_row in enumerate(target_data_reader):
if target_row_idx == 0:
self.stat_fieldnames, self.non_stat_fieldnames = self.extract_stat_fieldnames(
target_data_reader.fieldnames)
self.stat_fieldnames_set = set(self.stat_fieldnames)
self.run_logger.info(
"Scoring target data {}".format(target_row_idx + 1))
target_data_vector = self.extract_stats_data_vector_from_csv_row(
target_row)
posteriors_filepath = os.path.join(
output_directory,
"{}.posterior.{:03d}.samples{}.tsv".format(
output_name_prefix, target_row_idx + 1, output_suffix))
self.accept_reject(target_data_vector=target_data_vector,
priors_data_filepaths=priors_data_filepaths,
output_filepath=posteriors_filepath)
if self.is_output_target_params:
target_params_filepath = os.path.join(
output_directory,
"{}.posterior.{:03d}.target{}.tsv".format(
output_name_prefix, target_row_idx + 1,
output_suffix))
with open(target_params_filepath, "w") as target_params_f:
target_params_f.write(
self.field_delimiter.join(
self.non_stat_fieldnames))
target_params_f.write("\n")
target_params_f.write(
self.field_delimiter.join(
str(target_row[k])
for k in self.non_stat_fieldnames))
target_params_f.write("\n")
def _parse_deme_site_frequencies(self, filepath, field_name_prefix,
is_normalize_by_site_counts, lineage_pair,
locus_definition, results_d):
with utility.universal_open(filepath) as src:
lines = src.read().split("\n")
assert len(lines) == 4 and lines[3] == ""
header_row = lines[1].split("\t")
results_d_row = lines[2].split("\t")
assert len(header_row) == len(results_d_row)
for key, val in zip(header_row, results_d_row):
if not val:
continue
key = "{}.{}".format(field_name_prefix, key)
val = float(val)
if is_normalize_by_site_counts:
val = val / locus_definition.num_sites
results_d[key] = val
return results_d
def write_results(self):
for file_idx, file_info in enumerate(self.file_infos):
output_filepath = self.compose_output_path_f(
file_info.filepath, file_idx)
self.run_logger.info("Writing file {} of {}: '{}'".format(
file_idx + 1, len(self.file_infos), output_filepath))
with utility.universal_open(output_filepath, "w") as dest:
writer = utility.get_csv_writer(
dest=dest,
fieldnames=file_info.fieldnames,
delimiter=self.field_delimiter,
restval=self.missing_data_value,
)
writer.writeheader()
for data_row_idx in range(*file_info.data_row_idx_range):
if self.logging_frequency and data_row_idx > 0 and (
data_row_idx % self.logging_frequency) == 0:
self.run_logger.info(
"- Writing row {}".format(data_row_idx + 1))
row = {}
for field_name in file_info.fieldnames:
row[field_name] = self.fields[field_name][data_row_idx]
writer.writerow(row)
def accept_reject(self, target_data_vector, priors_data_filepaths,
output_filepath):
if self.rejection_criteria_type == "num":
num_to_retain = self.rejection_criteria_value
else:
num_to_retain = None
dest = utility.universal_open(output_filepath, "w")
all_prior_fieldnames = []
all_prior_fieldnames_set = None
accepted_heap = []
for fidx, priors_data_filepath in enumerate(priors_data_filepaths):
self.run_logger.info(
"Reading simulation file {} of {}: '{}'".format(
fidx + 1, len(priors_data_filepaths),
priors_data_filepath))
with utility.universal_open(priors_data_filepath) as src:
priors_data_reader = csv.DictReader(
src,
delimiter=self.field_delimiter,
quoting=csv.QUOTE_NONE)
for row_idx, row in enumerate(priors_data_reader):
if self.logging_frequency and row_idx > 0 and row_idx % self.logging_frequency == 0:
self.run_logger.info(
"Reading simulation file {} of {}, row {}".format(
fidx + 1, len(priors_data_filepaths),
row_idx + 1))
if row_idx == 0:
if fidx == 0:
all_prior_fieldnames = list(
priors_data_reader.fieldnames)
all_prior_fieldnames_set = set(
all_prior_fieldnames)
current_file_stat_fieldnames = set(
self.extract_stat_fieldnames(
priors_data_reader.fieldnames)[0])
s1 = current_file_stat_fieldnames - self.stat_fieldnames_set
if s1:
raise ValueError(
"File '{}': Following summary statistics fields not found in target: {}"
.format(priors_data_filepath,
", ".join(s1)))
s2 = self.stat_fieldnames_set - current_file_stat_fieldnames
if s2:
raise ValueError(
"File '{}': Following summary statistics fields given in target but not found here: {}"
.format(priors_data_filepath,
", ".join(s2)))
header_row = []
for fnidx, fn in enumerate(all_prior_fieldnames):
if self.is_write_summary_stats or fn not in self.stat_fieldnames_set:
header_row.append(fn)
if self.is_write_rejection_score:
header_row.append(
self.distance_score_fieldname)
header_row.append(
self.normalized_distance_score_fieldname)
dest.write("{}\n".format(
self.field_delimiter.join(header_row)))
else:
current_file_fieldnames = set(
priors_data_reader.fieldnames)
s1 = current_file_fieldnames - all_prior_fieldnames_set
if s1:
raise ValueError(
"File '{}': Following fields found, but not found in previous files: {}"
.format(priors_data_filepath,
", ".join(s1)))
s2 = all_prior_fieldnames_set - current_file_fieldnames
if s2:
raise ValueError(
"File '{}': Following fields found in previous files, but not found here: {}"
.format(priors_data_filepath,
", ".join(s2)))
try:
prior_data_vector = self.extract_stats_data_vector_from_csv_row(
row)
except SisterBayesRejectorStatsVectorValueException:
if self.is_ignore_invalid_priors_data_vectors:
continue
else:
raise
except SisterBayesRejectorStatsVectorSizeException:
if self.is_ignore_invalid_priors_data_vectors:
continue
else:
raise
distance_score = self.euclidean_distance(
target_data_vector, prior_data_vector)
row_values = self.field_delimiter.join(
row[fn] for fn in priors_data_reader.fieldnames
if self.is_write_summary_stats
or fn not in self.stat_fieldnames_set)
if self.is_write_rejection_score:
row_values = "{}{}{}".format(row_values,
self.field_delimiter,
distance_score)
# Normalize by number of comparisons
# How do we get this?
# Consider the following vectors:
# > x1 = c(3.1, 3.1, 3.1)
# > x2 = c(5.1, 5.1, 5.1)
# > y1 = c(3.1, 3.1, 3.1, 3.1, 3.1)
# > y2 = c(5.1, 5.1, 5.1, 5.1, 5.1)
# The naive/raw Euclidean distances are different, due to the different number of comparisons:
# > sqrt(sum((x2-x1)**2))
# [1] 3.464102
# > sqrt(sum((y2-y1)**2))
# [1] 4.472136
# But dividing the be sqrt of the length of the vectors makes them equal:
# > sqrt(sum((x2-x1)**2)) / sqrt(3)
# [1] 2
# > sqrt(sum((y2-y1)**2)) / sqrt(5)
# [1] 2
normalized_distance_score = distance_score / math.sqrt(
len(target_data_vector))
row_values = "{}{}{}".format(
row_values, self.field_delimiter,
normalized_distance_score)
heap_score = -1 * (distance_score)
heap_entry = (heap_score, row_values)
if self.rejection_criteria_type == "distance":
if distance_score <= self.rejection_criteria_value:
accepted_heap.append(heap_entry)
elif self.rejection_criteria_type == "num":
if len(accepted_heap) < num_to_retain:
accepted_heap.append(heap_entry)
if len(accepted_heap) == num_to_retain:
heapq.heapify(accepted_heap)
else:
heapq.heappushpop(accepted_heap, heap_entry)
else:
raise NotImplementedError(self.rejection_criteria_type)
# for fnidx, fn in enumerate(all_prior_fieldnames):
# value = row[fn]
# if self.is_write_summary_stats or fn not in self.stat_fieldnames_set:
# dest.write("{}{}".format(value, self.field_delimiter))
# dest.write("{}\n".format(distance))
accepted_heap.sort(reverse=True)
for hidx, heap_entry in enumerate(accepted_heap):
heap_entry = accepted_heap[hidx]
dest.write(heap_entry[1])
dest.write("\n")
dest.flush()
dest.close()
def main():
parser = argparse.ArgumentParser()
package_id = sisterbayes.package_id()
parser.add_argument("--version", action="version", version=package_id)
simulator_options = parser.add_argument_group("Simulation Configuration")
simulator_options.add_argument(
"configuration_filepath",
metavar="CONFIGURATION-FILE",
help="Path to file defining the simulation model and parameters.")
output_options = parser.add_argument_group("Output Options")
output_options.add_argument(
'-o',
'--output-name-prefix',
action='store',
dest='output_name_prefix',
type=str,
default=None,
metavar='NAME-PREFIX',
help=
"Prefix for output filenames (default: same as configuration filename stem)."
)
output_options.add_argument(
'-O',
'--output-directory',
action='store',
dest='output_directory',
type=str,
default=None,
metavar='DIRECTORY',
help="Directory for output files (default: current working directory)."
)
output_options.add_argument(
"-U",
"--unfolded-site-frequency-spectrum",
"--derived-site-frequency-spectrum",
action="store_true",
default=False,
help="Calculate the unfolded or derived site frequency spectrum."
" Otherwise, defaults to the folded or minor site frequency"
" spectrum.")
output_options.add_argument(
"--infinite-sites-model",
action="store_true",
default=False,
help="Use infinite sites model instead of finite sites.")
output_options.add_argument(
"--calculate-single-population-site-frequency-spectrum",
action="store_true",
default=False,
help="Calculate the single (within) population site frequency"
" spectrum in addition to the joint.")
output_options.add_argument(
"--concatenate-loci",
action="store_true",
default=False,
help="Concatenate statistics for all loci into one effective locus.",
)
output_options.add_argument(
"--no-normalize-by-concatenated-loci-count",
dest="is_normalize_by_concatenated_num_loci",
action="store_false",
default=True,
help=
"If concatenating loci, do NOT normalize frequency spectrum values by number of loci.",
)
output_options.add_argument(
"--no-normalize-by-site-counts",
dest="normalize_by_site_counts",
action="store_false",
default=True,
help=
"Do *not* normalize frequency spectrum values by number of sites in each locus."
)
output_options.add_argument(
"-l",
"--labels",
action="append",
help=
"Addition field/value pairs to add to the output (in format <FIELD-NAME>:value;)"
)
output_options.add_argument(
'--field-delimiter',
type=str,
default='\t',
help="Delimiter string separating fields in output (default: <TAB>').")
output_options.add_argument(
'--summary-stats-label-prefix',
type=str,
default='stat',
metavar='PREFIX',
help=
"Prefix for summar statistic field labels (default: '%(default)s').")
output_options.add_argument(
"--include-model-id-field",
action="store_true",
default=False,
help=
"Include a 'model.id' field (with same value as 'param.divTimeModel' field) in output."
)
output_options.add_argument(
"--append",
action="store_true",
default=False,
help="Append instead of overwriting output file(s).")
output_options.add_argument("--no-write-header",
action="store_true",
default=False,
help="Do not writer header row.")
output_options.add_argument("--raw-data",
action="store_true",
default=False,
help="Output raw data (alignments and trees).")
output_options.add_argument("--raw-data-alignment",
action="store_true",
default=False,
help="Output raw alignment.")
output_options.add_argument("--raw-data-mutation-tree",
action="store_true",
default=False,
help="Output raw mutation tree.")
output_options.add_argument("--raw-data-true-tree",
action="store_true",
default=False,
help="Output raw true tree.")
output_options.add_argument(
"--raw-data-alignment-format",
default="fasta",
choices=["fasta", "phylip", "nexus"],
help=
"Format for the raw data alignments ('fasta', 'phylip', or 'nexus'; default='fasta')."
)
output_options.add_argument(
"--raw-data-tree-format",
default="nexus",
choices=["nexus", "newick", "nexml"],
help=
"Format for the raw data trees ('nexus', 'newick', or 'nexml'; default='nexus')."
)
output_options.add_argument(
"--params-only-file",
action="store_true",
default=False,
help=
"Output file consisting of parameters only (for checking/validation).")
run_options = parser.add_argument_group("Run Options")
run_options.add_argument(
"-n",
"--num-reps",
type=int,
default=1,
help="Number of replicates (default: %(default)s).")
run_options.add_argument(
"-m",
"--num-processes",
default=1,
type=int,
help="Number of processes/CPU to run (default: %(default)s).")
run_options.add_argument("-z",
"--random-seed",
default=None,
help="Seed for random number generator engine.")
run_options.add_argument("-q",
"--quiet",
action="store_true",
help="Work silently.")
run_options.add_argument('--log-to-file',
action='store_true',
dest='log_to_file',
default=None,
help="Save log to file.")
run_options.add_argument(
"--log-frequency",
default=None,
type=int,
help=
"Frequency that background progress messages get written to the log (0: do not log informational messages)."
)
run_options.add_argument(
"--file-logging-level",
default="none",
choices=[
"debug",
"info",
"warning",
"error",
"critical",
"none",
],
help="Message level threshold for screen logs (default: %(default)s).")
run_options.add_argument(
"--stderr-logging-level",
default="info",
choices=[
"debug",
"info",
"warning",
"error",
"critical",
"none",
],
help="Message level threshold for screen logs (default: %(default)s).")
run_options.add_argument(
'-w',
'--working-directory-parent',
action='store',
type=str,
default=None,
help="Directory within which to create temporary directories and files."
)
run_options.add_argument("--no-cleanup",
action="store_true",
default=False,
help="Do not clean-up temporary files.")
run_options.add_argument("--debug-mode",
action="store_true",
default=False,
help="Run in debugging mode.")
fsc2_options = parser.add_argument_group("FastSimCoal2 Options")
fsc2_options.add_argument(
"--fsc2-path",
metavar="FSC2-PATH",
default=os.environ.get("SISTERBAYES_FSC2_PATH", "fsc"),
help="Path to FastsimCoal2 application (default: %(default)s).")
args = parser.parse_args()
config_d = {}
if not os.path.exists(args.configuration_filepath):
sys.exit("ERROR: Configuration file '{}' not found.".format(
args.configuration_filepath))
utility.parse_legacy_configuration(filepath=args.configuration_filepath,
config_d=config_d)
config_d["output_prefix"] = utility.output_prefix(
primary_source_filepath=args.configuration_filepath,
output_name_prefix=args.output_name_prefix,
output_directory=args.output_directory)
if args.log_frequency is None:
config_d["logging_frequency"] = int(args.num_reps / 10.0)
elif args.log_frequency == 0:
config_d["logging_frequency"] = None
else:
config_d["logging_frequency"] = args.log_frequency
config_d["fsc2_path"] = args.fsc2_path
if utility.which(config_d["fsc2_path"]) is None:
sys.exit("ERROR: FastSimCoal2 executable '{}' not found.\n"
"Install FastSimCoal2 and specify path to the executable\n"
"using the '--fsc2-path' argument.".format(
config_d["fsc2_path"]))
config_d["file_logging_level"] = args.file_logging_level
config_d["standard_error_logging_level"] = args.stderr_logging_level
config_d["log_to_file"] = args.log_to_file
config_d["log_to_stderr"] = not args.quiet
config_d[
"is_unfolded_site_frequency_spectrum"] = args.unfolded_site_frequency_spectrum
config_d[
"is_calculate_single_population_sfs"] = args.calculate_single_population_site_frequency_spectrum
config_d["is_calculate_joint_population_sfs"] = True
config_d["is_infinite_sites_model"] = args.infinite_sites_model
config_d["stat_label_prefix"] = args.summary_stats_label_prefix
config_d["supplemental_labels"] = utility.parse_fieldname_and_value(
args.labels)
config_d["field_delimiter"] = args.field_delimiter
config_d["is_include_model_id_field"] = args.include_model_id_field
config_d["is_concatenate_loci"] = args.concatenate_loci
config_d[
"is_normalize_by_concatenated_num_loci"] = args.is_normalize_by_concatenated_num_loci
config_d["is_normalize_by_site_counts"] = args.normalize_by_site_counts
is_store_raw_alignment = False
is_store_raw_mutation_tree = False
is_store_raw_true_tree = False
if args.raw_data:
is_store_raw_alignment = True
is_store_raw_mutation_tree = True
is_store_raw_true_tree = True
if args.raw_data_alignment:
is_store_raw_alignment = True
if args.raw_data_mutation_tree:
is_store_raw_mutation_tree = True
if args.raw_data_true_tree:
is_store_raw_true_tree = True
print(is_store_raw_alignment)
print(is_store_raw_mutation_tree)
print(is_store_raw_true_tree)
with utility.TemporaryDirectory(
prefix="sisterbayes-",
parent_dir=args.working_directory_parent,
is_suppress_cleanup=args.no_cleanup) as working_directory:
config_d["working_directory"] = working_directory
simulator = simulate.SisterBayesSimulator(
config_d=config_d,
num_processes=args.num_processes,
is_verbose_setup=True,
package_id=package_id,
is_store_raw_alignment=is_store_raw_alignment,
is_store_raw_mutation_tree=is_store_raw_mutation_tree,
is_store_raw_true_tree=is_store_raw_true_tree,
raw_data_output_prefix=config_d["output_prefix"],
raw_data_alignment_format=args.raw_data_alignment_format,
raw_data_tree_format=args.raw_data_tree_format,
is_debug_mode=args.debug_mode,
)
main_dest_filepath = config_d["output_prefix"] + ".stats.tsv"
dest = utility.universal_open(main_dest_filepath,
"a" if args.append else "w")
if args.params_only_file:
params_only_dest_filepath = config_d[
"output_prefix"] + ".params.tsv"
params_only_dest = utility.universal_open(
params_only_dest_filepath, "a" if args.append else "w")
else:
params_only_dest = None
# dest = utility.open_destput_file_for_csv_writer(
# filepath=filepath,
# is_append=args.append)
if args.append or args.no_write_header:
is_write_header = False
else:
is_write_header = True
with dest:
# writer = utility.get_csv_writer(
# dest=dest,
# delimiter=args.field_delimiter)
try:
results = simulator.execute(
nreps=args.num_reps,
dest=dest,
results_store=None,
params_only_dest=params_only_dest,
is_write_header=is_write_header,
)
except Exception as e:
sys.stderr.write(
"Traceback (most recent call last):\n {}{}\n".format(
" ".join(traceback.format_tb(sys.exc_info()[2])), e))
sys.exit(1)
if params_only_dest:
params_only_dest.close()
def main():
parser = argparse.ArgumentParser()
package_id = sisterbayes.package_id()
parser.add_argument("--version", action="version", version=package_id)
simulator_options = parser.add_argument_group("Configuration")
simulator_options.add_argument(
"configuration_filepath",
metavar="CONFIGURATION-FILE",
help="Path to the configuration file listing the data.")
processing_options = parser.add_argument_group("Processing Options")
processing_options.add_argument(
"-U",
"--unfolded-site-frequency-spectrum",
"--derived-site-frequency-spectrum",
action="store_true",
default=False,
help="Calculate the unfolded or derived site frequency spectrum."
" Otherwise, defaults to the folded or minor site frequency"
" spectrum.")
processing_options.add_argument(
"--calculate-single-population-site-frequency-spectrum",
action="store_true",
default=False,
help="Calculate the single (within) population site frequency"
" spectrum in addition to the joint.")
processing_options.add_argument(
"--no-normalize-by-site-counts",
dest="normalize_by_site_counts",
action="store_false",
default=True,
help=
"Do *not* normalize frequency spectrum by number of sites in each locus."
)
processing_options.add_argument(
"--concatenate-loci",
action="store_true",
default=False,
help="Collapse all loci and treat as a single locus for calculation.")
processing_options.add_argument(
"--concatenated-locus-label",
default=None,
help="If concatenating, label for the concatenated locus.")
output_options = parser.add_argument_group("Output Options")
output_options.add_argument(
'-o',
'--output-name-prefix',
action='store',
dest='output_name_prefix',
type=str,
default=None,
metavar='NAME-PREFIX',
help=
"Prefix for output filenames (default: same as configuration filename stem)."
)
output_options.add_argument(
'-O',
'--output-directory',
action='store',
dest='output_directory',
type=str,
default=None,
metavar='DIRECTORY',
help="Directory for output files (default: current working directory)."
)
output_options.add_argument(
"-l",
"--labels",
action="append",
help=
"Addition field/value pairs to add to the output (in format <FIELD-NAME>:value;)"
)
output_options.add_argument(
'--field-delimiter',
type=str,
default='\t',
help="Delimiter string separating fields in output (default: <TAB>').")
output_options.add_argument(
'--summary-stats-label-prefix',
type=str,
default='stat',
metavar='PREFIX',
help=
"Prefix for summar statistic field labels (default: '%(default)s').")
output_options.add_argument(
"--append",
action="store_true",
default=False,
help="Append instead of overwriting output file(s).")
output_options.add_argument("--no-write-header",
action="store_true",
default=False,
help="Do not writer header row.")
args = parser.parse_args()
config_d = utility.CaseInsensitiveDict()
utility.parse_legacy_configuration(filepath=args.configuration_filepath,
config_d=config_d)
config_d["output_prefix"] = utility.output_prefix(
primary_source_filepath=args.configuration_filepath,
output_name_prefix=args.output_name_prefix,
output_directory=args.output_directory)
config_d[
"is_unfolded_site_frequency_spectrum"] = args.unfolded_site_frequency_spectrum
config_d[
"is_calculate_single_population_sfs"] = args.calculate_single_population_site_frequency_spectrum
config_d["is_calculate_joint_population_sfs"] = True
config_d["stat_label_prefix"] = args.summary_stats_label_prefix
config_d["supplemental_labels"] = utility.parse_fieldname_and_value(
args.labels)
config_d["alignment_directory_head"] = os.path.dirname(
os.path.abspath(args.configuration_filepath))
config_d["field_delimiter"] = args.field_delimiter
config_d["is_concatenate_loci"] = args.concatenate_loci
config_d["concatenated_locus_label"] = args.concatenated_locus_label
config_d["is_normalize"] = args.normalize_by_site_counts
sscalc = sumstats.SisterBayesSummaryStatsCalculator(**config_d)
filepath = config_d["output_prefix"] + ".obs.sumstats.tsv"
# dest = utility.open_destput_file_for_csv_writer(
# filepath=filepath,
# is_append=args.append)
dest = utility.universal_open(filepath, "a" if args.append else "w")
if args.append or args.no_write_header:
is_write_header = False
else:
is_write_header = True
with dest:
# writer = utility.get_csv_writer(
# dest=dest,
# delimiter=args.field_delimiter)
try:
results = sscalc.write_summary_stats(
dest=dest, results_store=None, is_write_header=is_write_header)
except Exception as e:
sys.stderr.write(
"Traceback (most recent call last):\n {}{}\n".format(
" ".join(traceback.format_tb(sys.exc_info()[2])), e))
sys.exit(1)
def summarize(
self,
target_data_filepath,
):
if self.output_name_prefix is None:
self.output_name_prefix = os.path.splitext(
os.path.basename(target_data_filepath))[0]
if self.output_directory is None:
self.output_directory = "."
self.output_directory = os.path.realpath(self.output_directory)
output_prefix = os.path.join(self.output_directory,
self.output_name_prefix)
with utility.universal_open(target_data_filepath) as src:
reader = csv.DictReader(src,
delimiter=self.field_delimiter,
quoting=csv.QUOTE_NONE)
categorical_params = collections.OrderedDict()
continuous_params = collections.OrderedDict()
realized_div_time_samples = []
all_div_times = []
sp_labels = []
for row_idx, row in enumerate(reader):
realized_div_time_samples.append({})
for key_idx, key in enumerate(reader.fieldnames):
if key in categorical_params:
categorical_params[key][row[key]] += 1
elif key in continuous_params:
continuous_params[key].append(float(row[key]))
else:
if key in ("param.divTimeModel", "param.numDivTimes"):
val = row[key]
is_categorical = True
else:
try:
val = float(row[key])
is_categorical = False
except ValueError:
val = row[key]
is_categorical = True
if is_categorical:
categorical_params[key] = collections.Counter()
categorical_params[key][val] += 1
else:
continuous_params[key] = [val]
if key.startswith("param.divTime."):
sp_label = key.replace("param.divTime.", "")
realized_div_time_samples[-1][
sp_label] = continuous_params[key][-1]
all_div_times.append(val)
if row_idx == 0:
sp_labels.append(sp_label)
### EXPERIMENTAL ###
# categorical_params["param.effectiveDivTimeModel"] = self.cluster_by_relative_difference_threshold(
# sp_labels=sp_labels,
# realized_div_time_samples=realized_div_time_samples,
# all_div_times=all_div_times,
# relative_difference_threshold=0.01)
if self.cluster_criteria is not None:
if self.cluster_criteria == "bin_size":
cluster_results = self.cluster_by_bin_size(
sp_labels=sp_labels,
realized_div_time_samples=realized_div_time_samples,
all_div_times=all_div_times,
bin_size=self.cluster_criteria_value)
elif self.cluster_criteria == "num_bins":
cluster_results = self.cluster_by_num_bins(
sp_labels=sp_labels,
realized_div_time_samples=realized_div_time_samples,
all_div_times=all_div_times,
num_bins=self.cluster_criteria_value)
elif self.cluster_criteria == "absolute_difference_threshold":
cluster_results = self.cluster_by_absolute_difference_threshold(
sp_labels=sp_labels,
realized_div_time_samples=realized_div_time_samples,
absolute_difference_threshold=self.
cluster_criteria_value)
elif self.cluster_criteria == "relative_difference_threshold":
cluster_results = self.cluster_by_relative_difference_threshold(
sp_labels=sp_labels,
realized_div_time_samples=realized_div_time_samples,
all_div_times=all_div_times,
relative_difference_threshold=self.
cluster_criteria_value)
else:
raise ValueError(
"Unrecognized cluster criteria: '{}'".format(
self.cluster_criteria))
categorical_params[
"param.effectiveDivTimeModel"] = cluster_results
### EXPERIMENTAL ###
with utility.universal_open(
output_prefix + ".summary.continuous.tsv", "w") as dest:
row_results = collections.OrderedDict()
for param_idx, param_name in enumerate(continuous_params):
values = continuous_params[param_name]
row_results["param"] = param_name
summary = calclib.summarize(values)
row_results["mean"] = summary["mean"]
row_results["var"] = summary["var"]
row_results["sd"] = summary["sd"]
row_results["min"] = summary["range"][0]
row_results["max"] = summary["range"][1]
try:
row_results["hpd5"] = summary["hpd95"][0]
row_results["hpd95"] = summary["hpd95"][1]
except TypeError:
row_results["hpd5"] = "NA"
row_results["hpd95"] = "NA"
try:
row_results["quant5"] = summary["quant_5_95"][0]
row_results["quant95"] = summary["quant_5_95"][1]
except TypeError:
row_results["quant5"] = "NA"
row_results["quant95"] = "NA"
if param_idx == 0:
dest.write(
self.field_delimiter.join(row_results.keys()) +
"\n")
dest.write(
self.field_delimiter.join(
"{}".format(v)
for v in row_results.values()) + "\n")
for param_idx, param_name in enumerate(categorical_params):
with utility.universal_open(
output_prefix + ".summary.{:02d}.{}.tsv".format(
param_idx + 1, param_name), "w") as dest:
param_counter = categorical_params[param_name]
total = float(sum(param_counter.values()))
for category_idx, (category_name,
category_count) in enumerate(
param_counter.most_common()):
row_results = collections.OrderedDict()
row_results["label"] = category_name
row_results["freq"] = category_count / total
row_results["count"] = category_count
if category_idx == 0:
dest.write(
self.field_delimiter.join(row_results.keys()) +
"\n")
dest.write(
self.field_delimiter.join(
"{}".format(v)
for v in row_results.values()) + "\n")
def read_files(self, filepaths):
for file_idx, filepath in enumerate(filepaths):
self.run_logger.info("Reading file {} of {}: '{}'".format(
file_idx + 1, len(filepaths), filepath))
with utility.universal_open(filepath) as src:
self._read_file(src)
