下载NR库pyNCBI_Nr.py

    1. ## ascp -T -l640M -i /home/linxz/.aspera/connect/etc/asperaweb_id_dsa.openssh anonftp@ftp.ncbi.nlm.nih.gov:/blast/db/v5/nr.00.tar.gz /media/linxz/Data/BioDB/NCBI/NR/
    3. from ftplib import FTP
    4. from tqdm import tqdm
    5. import os
    7. #获取指定目录文件名并返回列表函数
    8. def get_files(file_dir,ext1 = 'gz',ext2='md5'): 
    9.     L=[] 
    10.     for root, dirs, files in os.walk(file_dir):
    11.         for file in files:
    12.             if os.path.splitext(file)[1] == '.' + ext1:
    13.                 L.append(file)
    14.             elif os.path.splitext(file)[1] == '.' + ext2:
    15.                 L.append(file)
    16.     return L
    18. ftp = FTP('ftp.ncbi.nlm.nih.gov')     # connect to host, default port 
    19. ftp.login()                     # user anonymous, passwd anonymous@ 登录
    20. ftp.cwd('blast/db/v5')               # 切换到blast/db/v5文件夹
    21. blast_db_list = ftp.nlst() #获取当前目录文件名，生成字典
    23. ascp_list = []
    24. db = 'nr'
    25. out_dir = '/media/linxz/Data/BioDB/NCBI/NR/' #输出文件夹路径
    26. ascp_cmd = 'ascp -T -l640M -i /home/linxz/.aspera/connect/etc/asperaweb_id_dsa.openssh anonftp@ftp.ncbi.nlm.nih.gov:/blast/db/v5/'
    28. finsh_list = get_files(out_dir) #调用函数获取指定目录文件名
    30. #构建shell命令
    31. for file in blast_db_list:
    32.     if db == file[:len(db)]:
    33.         if file not in finsh_list:
    34.             ascp_cmd += file
    35.             ascp_cmd += ' ' + out_dir
    36.             ascp_list.append(ascp_cmd)
    37.             ascp_cmd = 'ascp -T -l640M -i /home/linxz/.aspera/connect/etc/asperaweb_id_dsa.openssh anonftp@ftp.ncbi.nlm.nih.gov:/blast/db/v5/'
    38.         else:
    39.             print (file + '已完成下载')
    40. print ('开始下载')
    41. while ascp_list != []:
    42.     for cmd in tqdm(ascp_list):
    43.         os.system(cmd)
    44.         ascp_list.remove(cmd)