def seperate_parents_and_children(feature_db, parent_types_to_lift):
c = feature_db.conn.cursor()
relations = [
list(feature) for feature in c.execute(
'''SELECT * FROM relations join features as a on
a.id = relations.parent join features as b on b.id = relations.child''')
if feature[0] != feature[1]
]
all_ids = [
list(feature)[0] for feature in c.execute('''SELECT * FROM features''')
]
all_children_ids = [relation[1] for relation in relations]
all_parent_ids = [relation[0] for relation in relations]
lowest_children = np.setdiff1d(all_ids, all_parent_ids)
highest_parents = np.setdiff1d(all_ids, all_children_ids)
intermediates = set(all_children_ids).intersection(set(all_parent_ids))
parent_dict, child_dict, intermediate_dict = {}, {}, {}
add_parents(parent_dict, child_dict, highest_parents, parent_types_to_lift,
feature_db)
add_children(parent_dict, child_dict, lowest_children, feature_db)
add_intermediates(intermediates, intermediate_dict, feature_db)
parent_order = liftoff_utils.find_parent_order([
parent for parent in list(parent_dict.values()) if parent is not None
])
ref_feature_hierarchy = feature_hierarchy.feature_hierarchy(
parent_dict, intermediate_dict, child_dict)
return ref_feature_hierarchy, parent_order
def seperate_parents_and_children(feature_db, parent_types_to_lift):
parent_dict, child_dict, intermediate_dict = {}, {}, {}
feature_types = feature_db.featuretypes()
parent_types, child_types, intermediate_types = find_feature_type_hierarchy(feature_types, feature_db)
filtered_parents = [parent for parent in parent_types if parent in parent_types_to_lift]
for parent_type in filtered_parents:
for parent in feature_db.features_of_type(featuretype=parent_type):
add_parent(parent_dict, parent, child_dict)
add_children(child_types, parent, feature_db, child_dict)
add_intermediates(intermediate_types, intermediate_dict, feature_db)
parent_order = liftoff_utils.find_parent_order(
[parent for parent in list(parent_dict.values()) if parent is not None])
ref_feature_hierarchy = feature_hierarchy.feature_hierarchy(parent_dict, intermediate_dict, child_dict)
return ref_feature_hierarchy, parent_order
def check_homologues(all_lifted_features, lifted_features_to_check,
parent_dict, ref_parent_order, inter):
all_feature_list = liftoff_utils.get_parent_list(all_lifted_features)
features_to_check_list = liftoff_utils.get_parent_list(
lifted_features_to_check)
target_parent_order = liftoff_utils.find_parent_order(all_feature_list)
remap_features = {}
if inter == None:
inter = build_interval_list(all_feature_list)
for feature in features_to_check_list:
overlaps = find_overlaps(feature.start, feature.end, feature.seqid,
feature.strand,
feature.attributes["copy_id"][0], inter)
for overlap in overlaps:
feature_to_compare = overlap[2][1]
compare_overlapping_feature(feature_to_compare, parent_dict,
feature, remap_features,
ref_parent_order, target_parent_order)
return remap_features
def check_homologues(all_lifted_features, lifted_features_to_check,
parent_dict, ref_parent_order, max_overlap):
all_feature_list = liftoff_utils.get_parent_list(all_lifted_features)
features_to_check_list = liftoff_utils.get_parent_list(
lifted_features_to_check)
target_parent_order = liftoff_utils.find_parent_order(all_feature_list)
remap_features = set()
feature_locations = build_interval_list(all_feature_list)
for feature in features_to_check_list:
overlaps = liftoff_utils.find_overlaps(
feature.start - 1, feature.end - 1, feature.seqid, feature.strand,
feature.attributes["copy_id"][0], feature_locations, parent_dict,
all_lifted_features, max_overlap)
for overlap in overlaps:
if overlap[2][0] != feature.attributes["copy_id"][0]:
feature_to_compare = overlap[2][1]
compare_overlapping_feature(feature_to_compare, feature,
remap_features, ref_parent_order,
target_parent_order)
return remap_features, feature_locations
def check_homologues(all_lifted_features, lifted_features_to_check,
parent_dict, original_parent_order):
all_feature_list = liftoff_utils.get_parent_list(all_lifted_features,
parent_dict)
features_to_check_list = liftoff_utils.get_parent_list(
lifted_features_to_check, parent_dict)
all_feature_list.sort(key=lambda x: (x.seqid, x.start))
new_parent_order = liftoff_utils.find_parent_order(all_feature_list)
chrom_index_dict = {}
feature_index = 0
for feature in all_feature_list:
if feature.seqid not in chrom_index_dict:
chrom_index_dict[feature.seqid] = feature_index
feature_index += 1
remap_features = {}
for feature in features_to_check_list:
nearby_features = all_feature_list[chrom_index_dict[feature.seqid]:]
compare_nearby_features(nearby_features, parent_dict, feature,
remap_features, original_parent_order,
new_parent_order)
return remap_features