binary_sample.py

#!/usr/bin/env python3

import sys, os

samples = 100
sample_size = 25 * 1024 * 1024

with open(sys.argv[1], 'rb') as input, open(sys.argv[2], 'w+b') as output:
    total_size = os.fstat(input.fileno()).st_size
    sampled_size = samples * sample_size
    excluded_size = total_size - sampled_size
    intron_size = int(excluded_size / (samples - 1))

    for i in range(samples):
        sample = input.read(sample_size)
        output.write(sample)
        input.seek(int(intron_size), os.SEEK_CUR)