def save(self, file_path):
"""Save model to a file
"""
from bpnet.utils import write_pkl, SerializableLock
# fix the serialization of _OPERATIVE_CONFIG_LOCK
gin.config._OPERATIVE_CONFIG_LOCK = SerializableLock()
write_pkl(self, file_path)
def modisco_table(modisco_dir,
contrib_scores,
output_dir,
report_url=None,
contribsf=None,
footprint_width=200):
"""Write the pattern table to as .html and .csv
"""
plt.switch_backend('agg')
from bpnet.modisco.table import ModiscoData, modisco_table, write_modisco_table
from bpnet.modisco.motif_clustering import hirearchically_reorder_table
add_file_logging(output_dir, logger, 'modisco-table')
print("Loading required data")
data = ModiscoData.load(modisco_dir,
contrib_scores,
contribsf=contribsf,
footprint_width=footprint_width)
print("Generating the table")
df = modisco_table(data)
print("Writing the results")
write_modisco_table(df, output_dir, report_url, 'pattern_table')
print("Writing clustered table")
write_modisco_table(hirearchically_reorder_table(df, data.tasks),
output_dir, report_url, 'pattern_table.sorted')
print("Writing footprints")
profiles = OrderedDict([(pattern, {
task: data.get_profile_wide(pattern, task).mean(axis=0)
for task in data.tasks
}) for pattern in data.mf.pattern_names()])
write_pkl(profiles, Path(output_dir) / 'footprints.pkl')
print("Done!")
def modisco_export_patterns(modisco_dir, output_file, contribsf=None):
"""Export patterns to a pkl file. Don't cluster them
Adds `stacked_seqlet_contrib` and `n_seqlets` to pattern `attrs`
Args:
modisco_dir: modisco directory containing
output_file: output file path for patterns.pkl
"""
from bpnet.cli.contrib import ContribFile
logger.info("Loading patterns")
modisco_dir = Path(modisco_dir)
mf = ModiscoFile(modisco_dir / 'modisco.h5')
patterns = [mf.get_pattern(pname) for pname in mf.pattern_names()]
if contribsf is None:
contrib_file = ContribFile.from_modisco_dir(modisco_dir)
logger.info("Loading ContribFile into memory")
contrib_file.cache()
else:
logger.info("Using the provided ContribFile")
contrib_file = contribsf
logger.info("Extracting profile and contribution scores")
extended_patterns = []
for p in tqdm(patterns):
p = p.copy()
# get seqlets
valid_seqlets = mf._get_seqlets(p.name)
# extract the contribution scores
sti = contrib_file.extract(valid_seqlets, profile_width=None)
sti.dfi = mf.get_seqlet_intervals(p.name, as_df=True)
p.attrs['stacked_seqlet_contrib'] = sti
p.attrs['n_seqlets'] = mf.n_seqlets(p.name)
extended_patterns.append(p)
write_pkl(extended_patterns, output_file)