def filter_trf_gff_by_exact_copy_number(input_gff,
output_gff,
filtered_out_gff,
min_copy_number,
perfect_tandem=False):
if perfect_tandem:
def filtering_expression(gff_description_dict):
if (gff_description_dict["Pattern"] *
min_copy_number) in gff_description_dict["seq"]:
return True
return False
else:
def filtering_expression(gff_description_dict):
if gff_description_dict["seq"].count(
gff_description_dict["Pattern"]) >= min_copy_number:
return True
return False
AnnotationsRoutines.filter_gff_by_description(input_gff, output_gff,
filtered_out_gff,
filtering_expression)
def filter_trf_gff(self,
input_gff,
output_gff,
filtered_out_gff,
min_period=None,
max_period=None,
min_copy_number=None,
max_copy_number=None,
pattern=None,
min_percentage_of_matches=None,
max_percentage_of_indels=None,
min_entropy=None,
max_entropy=None):
def filtering_expression(gff_description_dict):
return self.gff_filtering_expression(
gff_description_dict,
min_period=min_period,
max_period=max_period,
min_copy_number=min_copy_number,
max_copy_number=max_copy_number,
pattern=pattern,
min_percentage_of_matches=min_percentage_of_matches,
max_percentage_of_indels=max_percentage_of_indels,
min_entropy=min_entropy,
max_entropy=max_entropy)
AnnotationsRoutines.filter_gff_by_description(input_gff, output_gff,
filtered_out_gff,
filtering_expression)