script

  1. import os
  2. import json
  3. import re
  5. class KOParser(object):
  6.     def __init__(self, map):
  7.         self.map = map
  9.     def json_parser(self):
  10.         # check the existance of json file
  11.         if os.path.exists(self.map):
  12.             with open(self.map, 'r') as f:
  13.                 ko_list = []
  14.                 # convert json to dict
  15.                 map_dict = json.load(f)
  16.                 # print(type(map_dict)) dict
  17.                 maps = map_dict['children']
  18.                 # print(type(maps)) list
  19.                 for map in maps:
  20.                     map_name = map['name'][0:5] + '\t' + map['name'][6:]
  21.                     # print(map_name)
  22.                     # print(type(map['children']))
  23.                     for map_l_1 in map['children']:
  24.                         map_l_1_name = map_l_1['name'][0:5] + '\t' + map_l_1['name'][6:]
  25.                         # print(map_l_1_name)
  26.                         for pathway in map_l_1['children']:
  27.                             try:
  28.                                 for genes in pathway['children']:
  29.                                     pathway_name = pathway['name'][0:5] + '\t' + pathway['name'][6:]
  30.                                     # print(genes['name'])
  31.                                     k_num = genes['name'].split(sep='  ')[0]
  32.                                     gene_name = genes['name'].split(sep='  ')[1].split(sep=';')[0]
  33.                                     anno = genes['name'].split(sep='  ')[1].split(sep=';')[-1]
  34.                                     try:
  35.                                         pattern = re.compile('(.*)(\[EC:.*\])')
  36.                                         product = re.search(pattern, anno).group(1)
  37.                                         ec = re.search(pattern, anno).group(2)
  38.                                         ko = k_num + '\t' + gene_name + '\t' + product + '\t' + ec
  39.                                     except:
  40.                                         ko = k_num + '\t' + gene_name + '\t' + anno
  41.                                     # print(ko)
  42.                                     info = map_name + '\t' + map_l_1_name + '\t' + pathway_name + '\t' + ko
  43.                                     ko_list.append(info)
  44.                             except:
  45.                                 continue
  46.                 return ko_list
  47.         else:
  48.             print('Error: Json file does not exist.')
  50.     def save_data(self, result, file_name):
  51.         # convert list to string for writing as file
  52.         for info in result:
  53.             kos = ''.join(info) + '\n'
  54.             with open(file_name, 'a', encoding='utf-8') as f:
  55.                 f.write(kos)
  58. if __name__ == '__main__':
  59.     map = 'ko00001.json'
  60.     file_name = 'ko.txt'
  61.     ko_json = KOParser(map)
  62.     ko = ko_json.json_parser()
  63.     ko_json.save_data(ko, file_name)

download json

choose Download json.

run

  1. python pathway_annotation.py

Reference

  1. 解析KEGG json文件