>>13727298>crispr < 10 years ago>stem cell reprogramming 2012>generative models 5 years ago>transformers, attention models <10 years ago>graph-convolution models, which could generalize convolutional models to graphs, are in their infancy and show amazing promise, most expect a general graph-convolutional network to be introduced in the next 5 years>no real solution to finding new molecules 3D shape-fitting exist>We are like babies working with small-molecules for drugs because we can't do anything close to biochemistry, making crazy-ass complicated chiral moleculesI mean, we are in our infancy of the "golden age" of biotechnology and there is so much goddamn shit we haven't come close to.
