A script for NCBI SRA downloads.

ncbi_url.py

import argparse

default_prefix = "ftp.sra.ebi.ac.uk/vol1/fastq/"
ascp_prefix = "ftp.sra.ebi.ac.uk:/vol1/fastq/"
default_protocol = "ftp://"
default_suffix = "*.fastq.gz"
default_ascp_line = "ascp -QT -l 300m -i ~/.ssh/asperaweb_id_dsa.openssh anonftp@ftp.ncbi.nlm.nih.gov:{} {}"


def prepare_url(srr_name, prefix):
    dir_1 = srr_name[:6]
    url = ""
    num_digits = sum(s.isdigit() for s in srr_name)
    if num_digits == 6:
        url = prefix + dir_1 + "/" + srr_name + "/"
    elif num_digits == 7:
        dir_2 = "00" + srr_name[-1]
        url = prefix + dir_1 + "/" + dir_2 + "/" + srr_name + "/"
    elif num_digits == 8:
        dir_2 = "0" + srr_name[-2:]
        url = prefix + dir_1 + "/" + dir_2 + "/" + srr_name + "/"
    elif num_digits == 9:
        dir_2 = srr_name[-3:]
        url = prefix + dir_1 + "/" + dir_2 + "/" + srr_name + "/"
    return url


def prepare_path(srr_name, prefix=None):
    dir_1 = srr_name[:6]
    suffix = ".sra"
    url = dir_1 + "/" + srr_name + "/" + srr_name
    return "{}{}{}".format(prefix, url, suffix)


def ascp_line(srr_name, out_dir):
    path = prepare_path(srr_name, prefix="/sra/sra-instant/reads/ByRun/sra/SRR/")
    return default_ascp_line.format(path, out_dir)


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument('--SRR', help='SRR name to be parsed')
    parser.add_argument('--file', help='file containing SRR name to be parsed')
    parser.add_argument('--ascp', action='store_true')
    parser.add_argument('--out', help='output directory')
    args = parser.parse_args()

    suffix = default_suffix

    if args.ascp:
        if args.file:
            with open(args.file) as srr_file:
                for line in srr_file:
                    print(ascp_line(line.strip(), args.out))
        else:
            print(ascp_line(args.SRR, args.out))
    else:
        protocol = default_protocol
        prefix = default_prefix
        if args.file:
            with open(args.file) as srr_file:
                for line in srr_file:
                    url = prepare_url(line.strip(), prefix)
                    print("{}{}{}".format(protocol, url, suffix))
        else:
            url = prepare_url(args.SRR, prefix)
            print("{}{}{}".format(protocol, url, suffix))


if __name__ == '__main__':
    main()