Tox21数据集学习笔记

Toxicology Testing in the 21st Century (Tox21)
比赛地址链接：https://tripod.nih.gov/tox21/challenge/data.jsp#

TOX21 program介绍

In fact, more than 30 percent of promising pharmaceuticals have failed in human clinical trials because they are determined to be toxic despite promising pre-clinical studies in animal models (Nat Rev Drug Discov. 2004;3(8):711–715). Creating new methods for assessing chemical toxicity has the potential to improve how scientists evaluate environmental chemicals and develop new medicines.
The Toxicology in the 21st Century (Tox21) program, a federal collaboration involving NIH, the Environmental Protection Agency, and the Food and Drug Administration, is aimed at developing better toxicity assessment methods. The goal is to quickly and efficiently test whether certain chemical compounds have the potential to disrupt processes in the human body that may lead to adverse health effects.
predict more effectively how a collection of 10,000 compounds composed of environmental chemicals and approved drugs will affect human health and the environment.
目的：快速判断化合物是否会扰乱人体过程，出现不良反应等。
predict compounds’ interference in biochemical pathways using only chemical structure data

数据：约10000个化合物，包括环保化合物、一些上市药物等。
from nuclear receptor signaling and stress pathway assays run against Tox21’s 10,000-compound library (Tox21 10K) to build models and look for structure-activity relationships.


这些化合物对于不同的靶点有不同的毒理活性表现（12个毒理试验），约12000个数据。

数据处理

rdkit 、molvs 去重复并标准化

training从11000+，合并相同的化合物及其12种毒性测试结果，最后training的结果缩减到了8038个左右。

描述符寻找

chemdes