Imperial College London
Alternate

ProfessorRobertGlen

Faculty of Medicine, Department of Metabolism, Digestion and Reproduction

Chair in Computational Medicine
 
 
 
//

Contact

 

+44 (0)20 7594 7912r.glen Website

 
 
//

Location

 

362Sir Alexander Fleming BuildingSouth Kensington Campus

//

Summary

 

Publications

Citation

BibTex format

@article{Robinson:2020:10.1007/s10822-019-00274-0,
author = {Robinson, MC and Glen, RC and Lee, AA},
doi = {10.1007/s10822-019-00274-0},
journal = {Journal of Computer-Aided Molecular Design},
pages = {717--730},
title = {Validating the validation: reanalyzing a large-scale comparison of deep learning and machine learning models for bioactivity prediction},
url = {http://dx.doi.org/10.1007/s10822-019-00274-0},
volume = {34},
year = {2020}
}
Download

RIS format (EndNote, RefMan)

TY  - JOUR
AB  - Machine learning methods may have the potential to significantly accelerate drug discovery. However, the increasing rate of new methodological approaches being published in the literature raises the fundamental question of how models should be benchmarked and validated. We reanalyze the data generated by a recently published large-scale comparison of machine learning models for bioactivity prediction and arrive at a somewhat different conclusion. We show that the performance of support vector machines is competitive with that of deep learning methods. Additionally, using a series of numerical experiments, we question the relevance of area under the receiver operating characteristic curve as a metric in virtual screening. We further suggest that area under the precision-recall curve should be used in conjunction with the receiver operating characteristic curve. Our numerical experiments also highlight challenges in estimating the uncertainty in model performance via scaffold-split nested cross validation.
AU  - Robinson,MC
AU  - Glen,RC
AU  - Lee,AA
DO  - 10.1007/s10822-019-00274-0
EP  - 730
PY  - 2020///
SN  - 0920-654X
SP  - 717
TI  - Validating the validation: reanalyzing a large-scale comparison of deep learning and machine learning models for bioactivity prediction
T2  - Journal of Computer-Aided Molecular Design
UR  - http://dx.doi.org/10.1007/s10822-019-00274-0
UR  - https://www.ncbi.nlm.nih.gov/pubmed/31960253
UR  - http://hdl.handle.net/10044/1/76584
VL  - 34
ER  -
Download