Our work on PROTEUS has successfully established a robust foundation for AI-driven protein engineering. However, this is just the beginning. We have a clear and exciting path forward to enhance its capabilities, validate its real-world impact, and contribute lasting value to the scientific community. Our future efforts will be focused on the following key areas:
From Prediction to Production: Bridging the Wet Lab Gap
Our top priority is to complete the final, crucial validation loop of our computational designs. While our in silico results are highly promising, the ultimate test lies in the crucible of wet lab experiments. In close collaboration with our biology team, we will:
- Synthesize and test our top candidate sequences for key proteins like A4GRB6_PSEAI_Chen_2020 and GFP_AEQVI_Sarkisyan_2016.
- Perform rigorous experimental characterization using methods such as enzyme activity assays, thermal stability analysis (e.g., DSF), and fluorescence spectroscopy.
Scaling Up and Scaling Out: Advanced Models and Broader Applications
To stay at the cutting edge and maximize the platform's utility, we plan to continuously expand its core capabilities.
We will explore fine-tuning larger and more advanced base models, such as Progen2 and other new generative architectures, to unlock new potential in protein design tasks.
We will apply our workflow to a wider range of proteins relevant to the iGEM community and beyond, including:
- Enzymes for key metabolic pathways
- Antibody fragments with therapeutic potential
- Protein components for novel biosensors
Beyond Single Points: Tackling Multi-Point Mutations and Epistasis
We recognize that the future of protein engineering lies in understanding the complex interplay between multiple mutations (epistasis). To overcome the limitations of our current single-point scanning strategy, we will develop more sophisticated algorithms to:
Implement combinatorial search strategies based on the high-quality single-point mutation maps we have already generated.
Introduce iterative modification processes, where a new round of scanning is performed on the background of an already successful mutation to discover synergistic effects.
From a Tool to an Ecosystem: Software Enhancement and Open Source
Our ultimate goal is to create a tool that empowers the entire synthetic biology community. To achieve this, we will:
Continuously improve the PROTEUS web application, optimizing the user experience and computational efficiency based on community feedback.
Release our core algorithms and trained models after the project concludes, accompanied by detailed technical documentation. We aim to contribute a powerful, transparent, and open computational tool that others can build upon.