【Python实现】解析Drugbank文件中的XML
1 样例一：xml直接提取
在做Drugbank数据处理时，发现的一个能够解决我问题的开源代码：项目地址以及Test文件，都在：
https://github.com/Deshan-Zhou/deal_DrugBank
但是直接上手处理xml格式的文件，对于xml数据的组织形式不熟悉的同志来讲很头疼，并不能够随心所欲的提取自己想提取的各种关联信息。在样例二中我提供一种能够方便提取数据的方法。
from xml.sax.handler import ContentHandler
from xml.sax import parse
import pandas as pd
"""
简写：
dbid : DrugBank id
dbname : DrugBank name
chid : ChEMBL id
ptid : protein id
"""
class ExtractData(ContentHandler):
def __init__(self):
#各个简写的映射关系
self.dbid_chid = {}
self.dbid_dbname = {}
self.dbid_dbid = {}
self.dbid_ptid = {}
#当前的药物id和遍历区域的限定
self.curr_id = ""
self.limit = 0
#可以自动获取遍历到的元素里面的内容，如<ele>content.....</ele>
def characters(self,content):
if self.limit == 2:
self.curr_id = content
self.limit = 3
elif self.limit == 4:
self.dbid_dbname[self.curr_id] = content
self.limit = 0
elif self.limit == 6:
self.dbid_dbid.setdefault(self.curr_id,set()).add(content)
self.limit = 5
elif self.limit == 8:
if content == "ChEMBL":
self.limit = 9
elif self.limit == 10:
self.dbid_chid[self.curr_id] = content
self.limit = 0
#遍历到标签开始时调用
def startElement(self,name,attrs):
if name == "drug":
self.limit = 1
if self.limit == 1 and name == "drugbank-id" and attrs:
if attrs["primary"] == "true":
self.limit = 2
elif self.limit == 3 and name=="name":
self.limit = 4
elif name == "drug-interactions":
self.limit = 5
elif self.limit == 5 and name == "drugbank-id":
self.limit = 6
elif name == "targets":
self.limit = 7
elif self.limit == 7 and name == "polypeptide":
self.dbid_ptid.setdefault(self.curr_id,set()).add(attrs["id"])
elif name == "resource" and self.limit!=7:
self.limit= 8
elif self.limit == 9 and name == "identifier":
self.limit = 10
#遍历到标签结束时调用
def endElement(self,name):
if name == "drug-interactions":
self.limit = 0
elif name == "targets":
self.limit = 0
#遍历结束时调用
def endDocument(self):
#DrugBank id和ChEMBL id的映射
list1_key=[]
list1_val=[]
list1_columns="ChEMBL_id",
for key,val in self.dbid_chid.items():
list1_key.append(key)
list1_val.append(val)
file1=pd.DataFrame(index=list1_key,columns=list1_columns,data=list1_val)
file1.to_csv('dbid_chid.csv')
#DrugBank id和drug name的映射
list2_key=[]
list2_val=[]
list2_columns="Drug_name",
for key,val in self.dbid_dbname.items():
list2_key.append(key)
list2_val.append(val)
file2=pd.DataFrame(index=list2_key,columns=list2_columns,data=list2_val)
file2.to_csv('dbid_dbname.csv')
#这个映射关系太多，后期再处理
# for data in self.dbid_dbid.items():
# print(3,data)
#DrugBank id和target id的相互作用映射
list4_key=[]
list4_val=[]
for key,val in self.dbid_ptid.items():
list4_key.append(key)
list4_val.append(list(val))
file4=pd.DataFrame(index=list4_key,data=list4_val)
file4.to_csv('dbid_ptid.csv')
parse('full database.xml',ExtractData())
2 样例二：xml转json格式
首先将xml数据转换成json格式（python的字典格式）的文件，这种处理方式不仅会减小整个drugbank 数据的大小，而且能够很方便的打开文件并且观察定位自己需要的信息，从而快速编写代码提取数据。
def xml_trans_json():
xml = open("data_process/full database.xml").read()
json_file = "data_process/data.json"
convertJson = xmltodict.parse(xml, encoding='utf-8')
jsonStr = json.dumps(convertJson, indent=1)
with open(json_file, 'w+', encoding='utf-8') as f:
f.write(jsonStr)
在后续的系列中，将会提取drug-ATC code ,drug-drug interactions 信息作为样例。
3 提取drug-target信息
def extract_drug_target():
with open('data_process/data.json', 'r') as f:
data = json.load(f)
drugs = data['drugbank']['drug']
data_list = []
for drug in drugs:
try:
drug_id = drug['drugbank-id'][0]['#text']
except:
drug_id = drug['drugbank-id']['#text']
drug_name = drug['name']
try:
targets = drug['targets']['target']
except:
continue
type = drug['@type']
if isinstance(targets, dict):
try:
uniprot_id = targets['polypeptide']['@id']
except:
continue
target_name = targets['name']
data_list.append([drug_id, drug_name, type, uniprot_id, target_name])
elif isinstance(targets, list):
for target in targets:
try:
uniprot_id = target['polypeptide']['@id']
except:
continue
target_name = target['name']
data_list.append([drug_id, drug_name, type, uniprot_id, target_name])
data = pd.DataFrame(data_list, columns=['drugbank_id',
'drug_name', 'type', 'uniprot_id', 'target_name'])
data.to_csv('data_process/drug_target.csv', index=False)