ProtST

ProtST is a pretraining framework designed for understanding and predicting protein sequences by integrating protein functions with biomedical texts. It supports various downstream tasks such as protein localization prediction and zero-shot protein classification, enhancing the capabilities of protein language models.