sv_type_stat.py

import argparse
from collections import defaultdict

from general_functions import get_sv_type, load_sv_conflict, sorted_barplot

def parse_args():
    parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)
    parser.add_argument("conflicts", type=str)
    parser.add_argument("--keep", "-k", type=str_list, default=None)
    parser.add_argument("--keep_file", "-f", type=str, default=None)
    parser.add_argument("--fig", type=str, default=None)

    args = parser.parse_args()
    return args

def str_list(s_list):
    return s_list.split(",")

if __name__ == "__main__":
    args = parse_args()
    conflict_file = open(args.conflicts)

    sv_types = defaultdict(lambda: 0)
    if args.keep_file:
        out_file = open(args.keep_file, "w")

    for conflict in conflict_file:
        svs = load_sv_conflict(conflict)

        local_types = defaultdict(lambda: 0)
        for sv in svs:
            sv_type = get_sv_type(sv)
            local_types[sv_type] += 1

        if len(local_types) > 1:
            sv_type = "COMBINATION"

        sv_types[sv_type] += 1

        if args.keep and sv_type in args.keep:
            out_file.write(f"{conflict}")


    if args.fig:
        plot = sorted_barplot(sv_types, reverse=False)
        plot.title("Frequency of SV conflicts by type")
        plot.xlabel("Frequency")
        plot.ylabel("Type")
        plot.savefig(args.fig, bbox_inches="tight")