OneProt
paper Your tags
Your notes
OneProt is a multimodal protein foundation model that aligns protein sequence, structure, binding-site and text modalities in a shared latent space via an ImageBind-style alignment (a frozen sequence encoder anchoring the other modalities). It supports retrieval and cross-modal transfer across protein representations without retraining per task.
From the Theis group's Helmholtz AI "PROFOUND" effort (joint→Helmholtz Munich); published in PLOS Computational Biology (2025).