==============================
We sought to understand if the base pre-trained ESM3 model has sufficient biological fidelity to generate functional proteins.
ESM3 stands for Evolutionary Scale Modeling 3, which is a deep learning model used in protein structure prediction. It is a type of language model that has been trained on large amounts of protein sequence data to predict the 3D structure of proteins.
In this study, we wanted to know if the ESM3 model could accurately predict the structure of proteins that have not been previously studied. We found that the model was able to generate functional proteins with high accuracy, suggesting that it has sufficient biological fidelity for this task.
This is an important finding because it could help accelerate the discovery of new proteins and their potential applications in medicine and biotechnology.
We set out to create a functional green fluorescent protein (GFP) with low sequence similarity to existing ones