顺序的fqlist拆解成sample fqsize fq格式 ，涉及正则取样本名

# -*- coding: utf-8 -*-
# @Time    : 2022/1/20 下午10:10
# @Author  : lizhichao
# @Email   : 13652030758@163.com
# @File    : sample_fqsize_fq.py
# @Software: PyCharm

import os,sys,re

def fqlist_new_fqsize(rawfqlist,output):
    p=re.compile('\S+\/(CK.*)\/\S+\/(\S+\.fq.gz)')

    with open(rawfqlist,"r") as raw,open(output,"w") as pu:


        content="Sample,FQ SIZE,FQ PATH\n"
        sample="CK28904380"
        fqcontent=""
        fqsize=0
        for line in raw:
            linepath=line.strip()
            if p.search(linepath):
                if str(p.search(linepath).group(1)) !=sample and fqsize>0:
                    content += "%s,%s%s\n" % (sample, round(fqsize, 2), fqcontent)
                    fqsize = 0
                    fqcontent = ""

                sample=str(p.search(linepath).group(1))

                fqcontent += ",%s" %linepath

                fqsize += (os.path.getsize(linepath)) / float(1024 * 1024 * 1024.0)

            else:
                print ("line not normal:%s"%line)

        content += "%s,%s%s\n" % (sample, round(fqsize, 2), fqcontent)
        pu.write(content)


if __name__=="__main__":
    rawfqlist=sys.argv[1]
    output=sys.argv[2]
    fqlist_new_fqsize(rawfqlist,output)

标签：__,fq,fqcontent,fqsize,sample,content,linepath
来源： https://www.cnblogs.com/koujiaodahan/p/15878471.html