Introduction
Former researchers from Meta AI, who developed an AI language model for biology, have recently launched a new startup called EvolutionaryScale. Led by Alexander Rives, the team aims to predict the structures of unknown proteins using their transformers-based model.
Funding and Valuation
The startup has successfully raised $40 million in seed financing, with Lux Capital leading the round. EvolutionaryScale is currently valued at $200 million. Notable AI investors Nat Friedman and Daniel Gross also participated in the funding.
The Importance of Predicting Protein Structures
Proteins play a crucial role in various biological processes and are often the targets of drugs to treat diseases. However, predicting the structure of a protein is complex due to the interactions between thousands of atoms within it. EvolutionaryScale’s model aims to solve this challenge by leveraging its vast database containing 700 million possible 3D protein structures.
Comparison to DeepMind’s AlphaFold
In 2020, DeepMind made headlines with the release of AlphaFold, an AI system capable of predicting protein structures. Despite its success, EvolutionaryScale’s model has demonstrated faster predictions, although with slightly lower accuracy. This suggests that AI in biology still has room for significant advancements.
Future Plans and Challenges
EvolutionaryScale projects a ten-year timeline for the application of AI models in biology. The startup plans to scale up its AI model and aims to integrate other biological data beyond protein structures. Additionally, the company envisions a general-purpose AI model for biology that can be used in various applications, such as developing programmable cells for medical purposes or designing molecular machines for environmental cleanup.
Competition and Potential Partnerships
EvolutionaryScale is not alone in its pursuit of AI applications in biology. Rival companies like Inceptive are exploring similar goals. However, EvolutionaryScale remains confident in its unique scaling hypothesis and believes that it can outperform existing models like AlphaFold. The startup also sees potential partnerships with companies like Isomorphic Labs, Insitro, and Recursion to enable advancements in drug discovery.
Conclusion
EvolutionaryScale, led by Alexander Rives, has secured substantial funding to advance its AI model for protein structure prediction. With its focus on scaling and integration with other biological data, the startup aims to revolutionize the field of biology and contribute to the development of new drugs, therapies, and biotech applications.