Tox21 dataset backup

Dataset backup for computational experiments in the paper:

"Ranking Molecules with Vanishing Kernels and a Single Parameter:
Applicability Domain Included"

Berenger, F. and Yamanishi, Y.

Data source

https://github.com/deepchem/deepchem/tree/master/datasets/tox21.csv

download date: 09/11/2018 at 13:59:05

Preparation protocol

All smiles strings (molecules) have been standardised using

https://github.com/flatkinson/standardiser

Molecules that did not pass standardisation have been removed. Cf. standardisation/errors.smi for such molecules.

All molecules tested on a given toxicity endpoint/target were copied into a specific directory for that target. All toxic molecules for a given target have had their name prefixed with the word "active". Each list of molecules was randomized.

Directory structure

tox21.csv: backup copy of the original data source

targets.txt: list of all toxicity endpoints in the dataset; one per line.
             Target names are in the same order than columns
             in the tox21.csv file.

TARGET/ligands_std_rand.smi: all toxic molecules for TARGET and all
                             non toxic molecules; in random order

standardisation/errors.smi: molecules that did not pass standardisation
standardisation/standardised.smi: molecules that passed standardisation

Bibliography

@article{Huang2016,
author = {Huang, Ruili and Xia, Menghang and Nguyen, Dac-Trung and Zhao, Tongan and Sakamuru, Srilatha and Zhao, Jinghua and Shahane, Sampada A. and Rossoshek, Anna and Simeonov, Anton},
title = {Tox21Challenge to Build Predictive Models of Nuclear Receptor and Stress Response Pathways as Mediated by Exposure to Environmental Chemicals and Drugs},
journal = {Frontiers in Environmental Science},
volume = {3},
pages = {85},
year = {2016},
doi = {10.3389/fenvs.2015.00085},
}