Summary of Bio2token: All-atom Tokenization Of Any Biomolecular Structure with Mamba, by Andrew Liu et al.
Bio2Token: All-atom tokenization of any biomolecular structure with Mamba
by Andrew Liu, Axel Elaldi, Nathan Russell, Olivia Viessmann
First submitted to arxiv on: 24 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 The paper proposes a novel approach to efficiently encoding and representing large 3D molecular structures at the atomic level, which is crucial for biomolecular design applications. A quantized auto-encoder architecture is developed to learn tokenizations of complete proteins, RNA, and small molecule structures with reconstruction accuracies below 1 Angstrom. The study demonstrates that a simple Mamba state space model achieves competitive accuracy compared to an SE(3)-invariant IPA architecture, while also scaling efficiently to systems with nearly 100,000 atoms. This work paves the way for the development of all-atom generative models that can learn from and generate complex molecular structures. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is about finding a better way to represent really big molecules in 3D space. Right now, most methods don’t try to be too accurate and instead use simpler approximations. The researchers developed a new approach called quantized auto-encoders that can accurately represent these large molecules at the atomic level. They tested their method on different types of molecules, including proteins and small molecules, and showed it works well even when dealing with really big structures. |
Keywords
* Artificial intelligence * Encoder
