Explore a 50-minute conference talk on HyperPCM, a novel approach to modeling drug-target interactions in drug discovery. Delve into the application of HyperNetworks for efficient information transfer between tasks during inference, enabling accurate predictions of drug-target interactions on unseen protein targets. Learn how this method achieves state-of-the-art performance on well-known benchmarks like Davis, DUD-E, and ChEMBL-derived datasets, particularly excelling in zero-shot inference scenarios. The talk covers background information, explains HyperNetworks and HyperPCM concepts, presents experimental results, and discusses benchmarking across bioactivity, kinase inhibition, and molecular docking. Conclude with key takeaways and participate in a Q&A session to deepen your understanding of this innovative approach in AI-driven drug discovery.
Overview
Syllabus
- Intro + Background
- Hypernetworks
- HyperPCM
- Experiments
- Benchmarking: Bioactivity
- Benchmarking: Kinase Inhibition
- Benchmarking: Molecular Docking
- Conclusions
- Q&A
Taught by
Valence Labs
Related Courses
-
Structured State-Space Sequence Models for De Novo Drug Design
-
Molecular Representations for Drug Discovery
-
MFBind - A Multi-Fidelity Approach for Evaluating Drugs in Generative Modeling
-
Conditional Graph Information Bottleneck for Molecular Relational Learning
-
Predicting Single-Cell Perturbation Responses for Unseen Drugs
-
Learning to Group Auxiliary Datasets for Molecule Prediction
