Summary of Structure-based Drug Design by Denoising Voxel Grids, By Pedro O. Pinheiro and Arian Jamasb and Omar Mahmood and Vishnu Sresht and Saeed Saremi
Structure-based drug design by denoising voxel grids
by Pedro O. Pinheiro, Arian Jamasb, Omar Mahmood, Vishnu Sresht, Saeed Saremi
First submitted to arxiv on: 7 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Biomolecules (q-bio.BM)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper presents VoxBind, a score-based generative model for 3D molecules conditioned on protein structures. The approach represents molecules as 3D atomic density grids and uses a 3D voxel-denoising network to learn and generate molecular structures. The neural empirical Bayes formalism is extended to the conditional setting, allowing for two-step generation: sampling noisy molecules from a Gaussian-smoothed conditional distribution using underdamped Langevin MCMC, followed by single-step denoising to estimate clean molecules. Compared to state-of-the-art models, VoxBind is simpler to train, faster to sample from, and achieves better results on extensive in silico benchmarks. The generated molecules are more diverse, exhibit fewer steric clashes, and bind with higher affinity to protein pockets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary VoxBind is a new way to create 3D molecule structures that are based on the shape of proteins. It works by first creating noisy versions of these molecules and then cleaning them up to make sure they are realistic. This approach does better than other methods at making molecules that are diverse, don’t clash with each other, and bind well to protein pockets. |
Keywords
» Artificial intelligence » Generative model
