Atomic AI
RNA drug discovery with atomic precision, combining AI foundation models and wet-lab assays to target RNA.
What is Atomic AI?
Atomic AI is a San Francisco biotechnology company that brings together machine learning and structural biology to discover RNA-targeted small molecules and RNA-based therapeutics. Its PARSE platform pairs deep learning models, including the ATOM-1 RNA foundation model, with purpose-built in-house wet-lab assays in a closed feedback loop. The company designs medicines aimed at traditionally undruggable RNA targets and partners with pharmaceutical firms to apply its platform. Its geometric deep learning work on RNA structure was featured in the journal Science.
Key Features
ATOM-1 foundation model
A large language model for RNA that predicts three-dimensional RNA structure and functional properties.
PARSE platform
A Platform for AI-driven RNA Structure Exploration that links deep learning models with in-house wet-lab assays.
Chemical mapping data integration
Uses experimental chemical mapping data to train models and optimise molecular design.
RNA-targeted small molecules
Designs selective and potent small molecules that bind difficult RNA targets.
Multi-modality RNA design
Supports RNA-based small molecules, mRNA vaccines, siRNA, and circular RNA therapeutics.
Wet-lab feedback loop
Combines computational predictions with laboratory validation to refine candidate molecules.
Pharma partnering
Offers collaborations that let partner companies apply the platform to their own target areas.
Pros & Cons
Advantages
- Built on peer-reviewed research, with geometric deep learning of RNA structure featured in the journal Science.
- Pairs AI predictions with in-house wet-lab assays rather than relying on computation alone.
- Addresses RNA targets that have historically been hard to drug, opening new therapeutic options.
- Backed by an interdisciplinary team and senior scientific advisors from industry and academia.
- Supports multiple RNA modalities, from small molecules to mRNA vaccines and siRNA.
Limitations
- This is a research-stage biotechnology company, not a self-serve software product you can sign up for.
- No public pricing, free trial, or product access is available; engagement is through direct partnering.
- Information on the platform is high level, with limited technical detail published publicly.
- Outputs are therapeutic programmes and partnerships rather than tools for general users.
Use Cases
Pharmaceutical companies seeking partners to discover RNA-targeted small molecules in their strategic areas.
Drug discovery teams aiming to pursue RNA targets previously considered undruggable.
Biotech collaborators wanting AI-driven predictions of RNA three-dimensional structure for therapeutic design.
Organisations developing RNA-based modalities such as mRNA vaccines, siRNA, or circular RNA therapeutics.
Research partners looking to combine machine learning models with experimental wet-lab validation.