OpenAI's CLIP model is quite impressive at connecting text with images, efficiently learning visual concepts from natural language supervision. Can we do the same for protein sequences and structure? Instead of text and images, we're integrating protein sequences with their structural information. ...