DeepDRK: a deep learning framework for drug repurposing through kernel-based multi-omics integration
Problem
Predicting how cancer cell lines respond to anticancer drugs is bottlenecked by the lack of effective methods for integrating multimodal, multisource omics data.
Why it matters
Precision cancer medicine depends on accurately matching patients to drugs, affecting treatment outcomes.
What the AI did
DeepDRK trained deep neural networks on over 20,000 pan-cancer cell line-anticancer drug pairs, using kernel-based similarity matrices integrating genomics, transcriptomics, epigenomics, chemical properties and drug-target interactions.
Results
Surpassed previous approaches with higher accuracy on benchmark cancer cell line datasets; specific metrics not stated in source.
Impact
Peer-reviewed framework offering improved drug response prediction for precision oncology research.
Record
| Venue | Briefings in Bioinformatics |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Yongcui Wang, Yingxi Yang, Shilong Chen, Jiguang Wang |
| Citations | 85 (OpenAlex, at import time) |
| References | 73 |
| Model or method | DeepDRK (deep neural networks with kernel-based multi-omics integration) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1093/bib/bbab048 |
| Open access | closed |
| Keywords | drug response, multi-omics integration, kernel methods, deep learning, cancer cell lines |
Importance rating
significant — thousands to millions affected
Precision cancer medicine bottleneck; model correlated with clinical outcomes and found repurposing candidates; abstract gives dataset size, not affected-population scale.