Today we’re doing a journal club. I'm joined by two of the authors of a paper at the intersection of Machine Learning and Computational Biology. The topic is proteins: the ubiquitous, diverse molecular machines that run our cells. We’ll be discussing a method to encode these proteins into a neural network’s representation of the world, aka its latent space, and looking for similar proteins in that encoding. This is essential RAG (Retrieval Augemented Generation) for proteins. They name this system TM-Vec, due to the TM-Score that serves as the input to the loss function.