Deep Learning Applications for Predicting Pharmacological Properties of Drugs and Drug Repurposing Using Transcriptomic Data
importance 5/10
ai assisted
Computational Drug Discovery Methods drug discovery stage: research
Problem
Predicting pharmacological properties and therapeutic uses of drugs, and repurposing existing drugs, is slow and costly.
Why it matters
Drug development is expensive, and identifying new uses for existing drugs could save time and lives.
What the AI did
A deep neural network trained on LINCS transcriptomic perturbation data (678 drugs, A549/MCF-7/PC-3 cell lines) classified drugs into 12 therapeutic categories, outperforming an SVM baseline.
Results
DNN outperformed SVM on every multiclass comparison; exact accuracy figures not stated in source
Impact
Evidence supports DNN classification superiority over SVM on this dataset only.
Record
| Venue | Molecular Pharmaceutics |
|---|---|
| Published | 2016 |
| Work type | article |
| Authors | Alexander M. Aliper, Sergey Plis, Artem V. Artemov, Alvaro Ulloa, Polina Mamoshina, Alex A. Zhavoronkov |
| Citations | 608 (OpenAlex, at import time) |
| References | 41 |
| Model or method | deep neural network (DNN) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1021/acs.molpharmaceut.6b00248 |
| Open access | hybrid |
| Keywords | drug repurposing, deep learning, transcriptomics, LINCS, pharmacology, classification |
Importance rating
5
real but narrower advance
Abstract frames this as proof of principle for drug classification/repurposing on 678 drugs; no scale, cost, or patient impact stated.