def join_multi_file(inputfile=None,
outputfile=None,
target_feature=None,
key_to_join=None,
matrix_files=()):
"""
Join attributes from a set of tabulated files.
"""
# -----------------------------------------------------------
# load the GTF
# -----------------------------------------------------------
gtf = GTF(inputfile, check_ensembl_format=False)
# -----------------------------------------------------------
# Check target feature
# -----------------------------------------------------------
feat_list = gtf.get_feature_list(nr=True)
if target_feature is not None:
target_feature_list = target_feature.split(",")
for i in target_feature_list:
if i not in feat_list + ["*"]:
message("Feature " + i + " not found.", type="ERROR")
else:
target_feature = ",".join(feat_list)
# -----------------------------------------------------------
# Do it
# -----------------------------------------------------------
for join_file in matrix_files:
gtf = gtf.add_attr_from_matrix_file(feat=target_feature,
key=key_to_join,
inputfile=join_file.name)
gtf.write(outputfile, gc_off=True)
gc.disable()
close_properly(outputfile, inputfile)
def join_attr(inputfile=None,
outputfile=None,
join_file=None,
has_header=False,
new_key=None,
target_feature=None,
key_to_join=None,
matrix=None):
"""
Join attributes from a tabulated file.
"""
# -----------------------------------------------------------
# Check argument consistency
# -----------------------------------------------------------
if matrix is True:
if new_key is not None:
message("--new-key and --matrix are mutually exclusive.",
type="ERROR")
else:
if new_key is None:
message("--new-key is required when --matrix is False.",
type="ERROR")
# -----------------------------------------------------------
# load the GTF
# -----------------------------------------------------------
gtf = GTF(inputfile, check_ensembl_format=False)
# -----------------------------------------------------------
# Check target feature
# -----------------------------------------------------------
feat_list = gtf.get_feature_list(nr=True)
if target_feature is not None:
target_feature_list = target_feature.split(",")
for i in target_feature_list:
if i not in feat_list + ["*"]:
message("Feature " + i + " not found.", type="ERROR")
else:
target_feature = ",".join(feat_list)
# -----------------------------------------------------------
# Do it
# -----------------------------------------------------------
if not matrix:
gtf = gtf.add_attr_from_file(feat=target_feature,
key=key_to_join,
new_key=new_key,
inputfile=join_file.name,
has_header=has_header)
gtf.write(outputfile, gc_off=True)
else:
gtf = gtf.add_attr_from_matrix_file(feat=target_feature,
key=key_to_join,
inputfile=join_file.name)
gtf.write(outputfile, gc_off=True)
gc.disable()
close_properly(outputfile, inputfile)
