miseqParser.py 文件源码

def get_all_fastq_files(data_dir):
    """
    recursively go down data_dir and get all fastq files

    arguments:
            data_dir -- the directory that has SampleSheet.csv in it

    return list containing path for fastq files
    """

    pattern = fn_translate("*.fastq.*")
    fastq_files_path = path.join(data_dir, "Data", "Intensities", "BaseCalls")

    try:
        file_list = listdir(fastq_files_path)
        fastq_file_list = [path.join(fastq_files_path, file)
                           for file in file_list if re.match(pattern, file)]
        fastq_file_list.sort()

    except OSError:
        msg = "Invalid directory " + fastq_files_path
        raise OSError(msg)

    return fastq_file_list