D vs Python

count_reads_per_sample.d

import std.stdio;
import std.file;
import std.path;
import std.process;
import std.algorithm;
import std.parallelism;

void main(string[] args) {
    int[string] no_of_reads;

    foreach (i, arg; taskPool.parallel(args[1..args.length])) {
        if (!arg.isFile()) continue;

        File file;
        if (arg[arg.length - 2..arg.length] == "gz") {
            auto pipe = pipeShell("gunzip -c " ~ arg);
            file = pipe.stdout;
        } else {
            file = File(arg);
        }

        auto basename = baseName(arg);
        no_of_reads[basename] = file.byLine().map!(a => 1).sum() / 4;
    }

    foreach (string key; sort(no_of_reads.keys)) {
        writefln("%s\t%d", key, no_of_reads[key]);
    }
}

count_reads_per_sample.py

#!/usr/bin/python

import gzip
import mmap
import sys
import os.path

def count_fastq_records(handle):
	return int(handle.read(-1).count('\n')/4)

def sample_name(filename):
	return '.'.join(os.path.basename(filename).split('.')[0:2])

samples = dict()
for filename in sys.argv[1:]:
	with open(filename, 'r') as f:
		handle = mmap.mmap(f.fileno(), 0, access=mmap.ACCESS_READ)
		if filename.endswith('.gz'):
			handle = gzip.GzipFile(mode='r', fileobj=handle)
		samples[sample_name(filename)] = count_fastq_records(handle)

for sample in sorted(samples.keys()):
	print '%s\t%d' % (sample, samples[sample])