利用含PUBCHEM_CID的的csv文件获取assaysummary数据

最新推荐文章于 2021-10-23 22:51:18 发布

贺俊宏

最新推荐文章于 2021-10-23 22:51:18 发布

阅读量1.1k

点赞数

分类专栏：药用数据库的数据提取与使用

本文链接：https://blog.csdn.net/recher_He1107/article/details/100705112

版权

药用数据库的数据提取与使用专栏收录该内容

18 篇文章 59 订阅

订阅专栏


import pandas as pd
import numpy as np
import os
import sys
import requests

def download(url,file_name):
    s = requests.Session()
    response = s.get(url,stream=True)
    with open(file_name,'ab') as f:
        for item in response.iter_content(chunk_size=512):
            if item:
                f.write(item)

def getCIDfromFILE(file):
    CID = pd.read_csv(file)['PUBCHEM_CID']
    CID = list(CID.dropna().apply(lambda x:int(x)))  # TO be sure CIDS are digits
    return CID  

def getASSAYSUMMARYfromCID(file):
    CID = getCIDfromFILE(file)
    url_list =['https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/cid/{}/assaysummary/CSV'.format(i) for i in CID]
    for i in url_list:
        download(i,'assaysummary_CSV') 
        print('{} finished'.format(i))# tracing weather procedure is running
    df = pd.read_csv('assaysummary_CSV',sep = ',')
    return df

df = getASSAYSUMMARYfromCID(file)
df=df[df.ix[:,'AID'].map(lambda x:x.isdigit())]# erase 404data and  "columns" of CIDS
df.to_csv(outcome_file.csv) 

#file and outcome_file should be modified by your environment