We co-authored a study introducing a new tool called TemStaPro, which uses deep learning and protein language models to predict protein thermostability from its sequences.
Accurate prediction of protein thermostability is crucial for academic and industrial research. Leveraging machine learning, particularly deep learning, we utilize transfer learning with protein language models (pLMs) to predict thermostability. Our method, TemStaPro, trained on millions of sequences, accurately predicts thermostability for CRISPR-Cas Class II effector proteins (C2EPs), aligning with experimental data.
Find the full publication here: https://doi.org/10.1093/bioinformatics/btae157.