ambiguity_distances.py

from itertools import combinations
import sys

import matplotlib.pyplot as plt

from general_functions import load_sv_conflict

def log_bins(base, end):
    res = []
    x = base
    while x < end:
        res.append(x)
        x *= base

    res.append(x)

    return res


if __name__ == '__main__':
    fn = sys.argv[1]
    method = sys.argv[2]
    direction_start = sys.argv[3] == "start"
    max_dist = int(sys.argv[4])

    f = open(fn, "r")
    distances = []
    for line in f:
        called_svs = load_sv_conflict(line)

        break_positions = []
        for sv in called_svs:
            if sv.startswith(method):
                info = sv.split("_")
                # info looks like ["svim", "DEL", "-133", "232424", "323242", ...]
                if direction_start:
                    break_positions.append(int(info[3]))
                else:
                    break_positions.append(int(info[4]))

        # iterate over all 2-combinations of the conflicting SV calls
        for a, b in combinations(break_positions, 2):
            if (dist := abs(a - b)) < max_dist:
                distances.append(dist)

    plt.hist(distances, log=True)#, bins=log_bins(2, max(distances)))
    plt.xlabel(f"{'Start' if direction_start else 'End'} breakpoint " +
                "distances between ambiguous SV calls in a region")
    plt.ylabel("Frequency")
    plt.title("Distribution of breakend distances between ambiguous SVs in a graph")
    plt.savefig(f"ambiguity_distance_distribution_{method}.png", bbox_inches="tight")