genome_prepare/script/uniprot_info.py

import os,sys
import re
import optparse
from  collections import defaultdict
import time

def Time():
    return time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())

def make_dir(dir):
    if not os.path.exists(dir):
        dir=os.path.expanduser(dir)
        os.makedirs(dir)
        
        
def uniport_info_get():
    uniprot_info={}
    with open("/Business/psn_company/t01/public/Database/Uniprot_info/swissprot.info.txt","r") as fi:
        for line in fi:
            uniprot_id=re.search(r"\|(.*?)\|",line).group(1)
            if re.search(r" GN=",line):
                gene_name=re.search(r" GN=(.*?) ",line).group(1)
                uniprot_info[uniprot_id]=gene_name
            else:
                uniprot_info[uniprot_id]="-"
    return uniprot_info

def new_Annotation_get(uniprot_info,Annotation,out_file):
    with open(f'{Annotation}',"r") as fi,open(out_file,"w") as fw:
        for line in fi:
            lines=line.strip().split("\t")
            gene_id=lines[0]
            swiss_id=lines[1]
            if re.search(r"^tr\|",swiss_id) or re.search(r"^sp\|",swiss_id):
                if re.search(r' GN=(.*?) ', swiss_id):
                    swiss_name =  re.search(r' GN=(.*?) ', swiss_id).group(1)
                    fw.write(gene_id+f"\t{swiss_name}\n")
                else:
                    fw.write(gene_id+f"\t-\n")
            elif swiss_id != "-":
                swiss_info = re.split(r'[;,:]', swiss_id)
                swiss_name=[]
                #保留唯一名字，这个名字必须是uniprot对应的Gene Name ，避免插入的Name 为Swissprot ID（针对有多个uniprotId处理）
                name_insert_first=0
                for info in swiss_info:
                    if info in uniprot_info :
                        if uniprot_info[info]  != "-":
                            if name_insert_first==0:
                                name_insert_first =1
                                swiss_name.insert(0,uniprot_info[info])
                            else:
                                swiss_name.append(uniprot_info[info])
                        else:
                            swiss_name.append(info)
                    else:
                        swiss_name.append(info)
                swiss_all_name=";".join(swiss_name)
                if len(swiss_name) >0 :
                    swiss_uniq_name=swiss_name[0]
                else:
                    swiss_uniq_name="-"
                fw.write(gene_id+f"\t{swiss_uniq_name}\n")
            else:
                fw.write(gene_id+f"\t-\n")
    return 1

if __name__ == "__main__":
    parser=optparse.OptionParser(usage='"usage:%prog [options] arg1,arg2"',version="%prog 1.2")  
    parser.add_option('-s','--Swissprot',  
                    action='store',dest='Swissprot',  
                    help='Swissprot file')
    parser.add_option('-o','--outdir',  
                    action='store',dest='outdir',default = './',  
                    help='output directory [default:%default]')
    
    options,args=parser.parse_args()  
    if not options.Swissprot:
        os.system("python3 "+sys.argv[0]+" -h")
        sys.exit(1)
    outdir=os.path.abspath(os.path.expanduser(options.outdir))
    make_dir(outdir)
    
    outfile=os.path.join(outdir, "SwissprotName")
    uniport_info=uniport_info_get()
    Annotation_get=new_Annotation_get(uniport_info,options.Swissprot,outfile)
    if Annotation_get == 0:
        os.system(f"cp {options.Swissprot} {outfile}")
        print("This Annotation  file is not exists,please check it!")