nanoQC.py 文件源码

def get_bin(fq, sizeRange):
    '''
    Loop over the fastq file
    Extract list of nucleotides and list of quality scores in tuples in list
    Only select those reads of which the length is within the size range
    '''
    logging.info("Extracting nucleotides and quality scores of selected bin.")
    return [(list(rec.seq), list(rec.letter_annotations["phred_quality"]))
            for rec in SeqIO.parse(fq, "fastq") if sizeRange[0] < len(rec) < sizeRange[1]]