FastqReadNames.py 文件源码

def match_both_pairs_filter(self):
        self.logger.warning("Extracting read names from FASTQ file where both reads must match")
        regex = re.compile(r'/[12]$')
        base_read_names = {}
        with open(self.fastq_file, "r") as fastq_file_input:
            for record in SeqIO.parse(fastq_file_input, "fastq"):
                base_read_name = regex.sub('', record.id)

                if base_read_name in base_read_names:
                    base_read_names[base_read_name] = record.id
                else:
                    base_read_names[base_read_name] = 1

        with open(self.output_readnames_file, "w") as readnames_output:
            for base_name, read_name in base_read_names.items():
                if read_name== 1:
                    continue
                readnames_output.write(read_name + '\n')