def map_extra_copies(target_fasta, reference_fasta, ref_chroms, target_chroms,
processes, lifted_feature_list, parent_dict,
children_dict, feature_db, intermediate_dict,
parent_order, seq_threshold, minimap2_path, inter_files,
remap, max_alns):
liftoff_utils.clear_scores(lifted_feature_list, parent_dict)
unmapped_features = []
liftover_type = "copies"
extract_features.get_gene_sequences(parent_dict, ref_chroms,
reference_fasta, processes,
inter_files, liftover_type)
aligned_segments = align_features.align_features_to_target(
ref_chroms, target_chroms, processes, target_fasta, parent_dict,
children_dict, liftover_type, unmapped_features, reference_fasta,
minimap2_path, inter_files, remap, max_alns)
print("lifting features")
lift_features.lift_all_features(aligned_segments, {}, 0.0, feature_db,
parent_dict, children_dict,
intermediate_dict, unmapped_features,
lifted_feature_list, seq_threshold)
fix_overlapping_features.fix_incorrectly_overlapping_features(
lifted_feature_list, lifted_feature_list, parent_dict,
aligned_segments, unmapped_features, 0.0, intermediate_dict,
children_dict, feature_db, parent_order, seq_threshold)
def map_unplaced_genes(unmapped_features, target_fasta, reference_fasta,
ref_chroms, target_chroms, processes,
lifted_feature_list, feature_db, parent_dict,
intermediate_dict, children_dict, parent_order,
minimap2_path, inter_files, max_alns):
liftoff_utils.clear_scores(lifted_feature_list, parent_dict)
liftover_type = "unplaced"
unplaced_dict = {}
for feature_name in parent_dict:
feature = parent_dict[feature_name]
if feature.seqid in ref_chroms:
unplaced_dict[feature.id] = feature
extract_features.get_gene_sequences(unplaced_dict, ref_chroms,
reference_fasta, processes,
inter_files, liftover_type)
aligned_segments = align_features.align_features_to_target(
ref_chroms, target_chroms, processes, target_fasta, unplaced_dict,
children_dict, liftover_type, unmapped_features, reference_fasta,
minimap2_path, inter_files, True, max_alns)
print("lifting features")
lift_features.lift_all_features(aligned_segments, {}, 0.0, feature_db,
unplaced_dict, children_dict,
intermediate_dict, unmapped_features,
lifted_feature_list, 0.0)
fix_overlapping_features.fix_incorrectly_overlapping_features(
lifted_feature_list, lifted_feature_list, parent_dict,
aligned_segments, unmapped_features, 0.0, intermediate_dict,
children_dict, feature_db, parent_order, 0.0)
def map_extra_copies(ref_chroms, target_chroms, lifted_features_list, feature_hierarchy, feature_db, ref_parent_order,
args):
liftoff_utils.clear_scores(lifted_features_list, feature_hierarchy.parents)
unmapped_features = []
liftover_type = "copies"
extract_features.get_gene_sequences(feature_hierarchy.parents, ref_chroms, args, liftover_type)
min_cov, min_seqid = 0, args.sc
align_and_lift_features(ref_chroms, target_chroms, args, feature_hierarchy, liftover_type, unmapped_features,
feature_db, lifted_features_list, ref_parent_order, min_cov, min_seqid, args.overlap)
def map_unplaced_genes(unmapped_features, ref_chroms, target_chroms,
lifted_features_list, feature_db, feature_hierarchy, ref_parent_order, args):
liftoff_utils.clear_scores(lifted_features_list, feature_hierarchy.parents)
liftover_type = "unplaced"
unplaced_dict = get_features_from_unplaced_seq(ref_chroms, feature_hierarchy)
extract_features.get_gene_sequences(unplaced_dict, ref_chroms, args, liftover_type)
if args.exclude_partial:
min_cov, min_seqid = args.a, args.s
else:
min_cov, min_seqid = 0.0, 0.0
align_and_lift_features(ref_chroms, target_chroms, args, feature_hierarchy, liftover_type, unmapped_features,
feature_db, unplaced_dict, lifted_features_list, ref_parent_order, min_cov, min_seqid)
def map_unmapped_genes_agaisnt_all(unmapped_features, ref_chroms, target_chroms, lifted_features_list, feature_db,
feature_hierarchy, ref_parent_order, args):
liftoff_utils.clear_scores(lifted_features_list, feature_hierarchy.parents)
unmapped_dict = get_unmapped_genes(unmapped_features)
if args.exclude_partial:
min_cov, min_seqid = args.a, args.s
else:
min_cov, min_seqid = 0.05, 0.05
liftover_type = "unmapped"
extract_features.get_gene_sequences(unmapped_dict, ref_chroms, args, liftover_type)
unmapped_features = []
align_and_lift_features(ref_chroms, target_chroms, args, feature_hierarchy, liftover_type, unmapped_features,
feature_db, lifted_features_list, ref_parent_order, min_cov, min_seqid, args.overlap,)
return unmapped_features