kraken_trawl.py 文件源码

def download_gbk(assemb_tab, cmd, outdir = '.'):
    '''
    This should take a list of accessions and paths, and produce a
    source/destination pairs file.

    It is then possible to use the --file-pair-file to download all at once,
    and we can add the following flags to the ascp command:

    --overwrite=diff -k2

    the -k2 means files are compared with sparse checksums.
    '''

    fo = open("aspera_assemblies_src_dest.txt", 'w')
    assembs_dic_list = [parse_aseemb_rows(row) for row in assemb_tab.itertuples()]
    for assembly in assembs_dic_list:
        fo.write("{}\n{}\n".format(assembly['source'], assembly['dest']))
    fo.close()

    cmd = cmd.format(outdir, 'aspera_assemblies_src_dest.txt', outdir)
    print("Running the aspera cmd: {}".format(cmd), file = sys.stderr)
    p = subprocess.Popen( shlex.split(cmd))
    p.communicate()
    files_to_check = [a['gbk_file'] for a in assembs_dic_list]
    was_updated = parse_aspera_manifest_file(outdir, files_to_check, "aspera_assemblies_manifest.txt")
    print("Finished downloading all genomes.", file = sys.stderr)
    return assembs_dic_list

### LOADING GENOMES TO THE KRAKEN STAGGING AREA ################################