def contains_child(aln, children_coords, parent):
for child_interval in children_coords:
relative_start = liftoff_utils.get_relative_child_coord(parent, child_interval[0], aln.is_reverse)
relative_end = liftoff_utils.get_relative_child_coord(parent, child_interval[1], aln.is_reverse)
child_start, child_end = min(relative_start, relative_end), max(relative_start, relative_end)
overlap = liftoff_utils.count_overlap(child_start, child_end, aln.query_block_start, aln.query_block_end)
if overlap > 0:
return True
return False
def get_node_weight(aln, children_coords, parent):
weight = 0
for child_interval in children_coords:
relative_start = liftoff_utils.get_relative_child_coord(
parent, child_interval[0], aln.is_reverse)
relative_end = liftoff_utils.get_relative_child_coord(
parent, child_interval[1], aln.is_reverse)
child_start, child_end = min(relative_start, relative_end), max(
relative_start, relative_end)
weight += len(aln.mismatches[(aln.mismatches >= child_start)
& (aln.mismatches <= child_end)])
return weight
def find_mismatched_bases(start, end, shortest_path_nodes, parent):
all_mismatches = []
relative_start = liftoff_utils.get_relative_child_coord(
parent, start, shortest_path_nodes[0].is_reverse)
relative_end = liftoff_utils.get_relative_child_coord(
parent, end, shortest_path_nodes[0].is_reverse)
for node in shortest_path_nodes:
node_mismatches = np.array(node.mismatches)
mismatches = node_mismatches[
np.where((node_mismatches >= relative_start)
& (node_mismatches <= relative_end))[0]]
if len(mismatches) > 0:
all_mismatches.extend(mismatches.tolist())
return all_mismatches
def convert_coord(coord, parent, shortest_path_nodes):
relative_coord = liftoff_utils.get_relative_child_coord(
parent, coord, shortest_path_nodes[1].is_reverse)
if len(shortest_path_nodes) == 2:
return 0
for i in range(1, len(shortest_path_nodes) - 1):
if relative_coord >= shortest_path_nodes[i].query_block_start:
if relative_coord <= shortest_path_nodes[i].query_block_end:
return shortest_path_nodes[i].reference_block_start + (
relative_coord - shortest_path_nodes[i].query_block_start)
elif relative_coord < shortest_path_nodes[i + 1].query_block_start:
return shortest_path_nodes[i].reference_block_end
return shortest_path_nodes[1].reference_block_start
def get_edge_weight(from_node, to_node, children_coords, parent):
unaligned_range = [
from_node.query_block_end + 1, to_node.query_block_start - 1
]
unaligned_exon_bases = 0
for child_interval in children_coords:
if from_node.reference_name == "start":
is_reverse = to_node.is_reverse
else:
is_reverse = from_node.is_reverse
relative_start = liftoff_utils.get_relative_child_coord(
parent, child_interval[0], is_reverse)
relative_end = liftoff_utils.get_relative_child_coord(
parent, child_interval[1], is_reverse)
child_start, child_end = min(relative_start, relative_end), max(
relative_start, relative_end)
overlap = liftoff_utils.count_overlap(child_start, child_end,
unaligned_range[0],
unaligned_range[1])
if overlap == 1 and unaligned_range[0] == unaligned_range[1]:
unaligned_exon_bases += to_node.reference_block_start - from_node.reference_block_end + 1
else:
unaligned_exon_bases += max(0, overlap)
return unaligned_exon_bases + 1
def convert_coord(coord_start, coord_end, parent, shortest_path_nodes):
lifted_coords = []
for i in [coord_start, coord_end]:
lifted_coord = 0
relative_coord = liftoff_utils.get_relative_child_coord(
parent, i, shortest_path_nodes[0].is_reverse)
for j in range(0, len(shortest_path_nodes)):
if relative_coord >= shortest_path_nodes[j].query_block_start:
if relative_coord <= shortest_path_nodes[j].query_block_end:
lifted_coord = (
shortest_path_nodes[j].reference_block_start +
(relative_coord -
shortest_path_nodes[j].query_block_start))
lifted_coords.append(lifted_coord)
return lifted_coords[0], lifted_coords[1]
