Decode, design, deliver: machine learning revolutionizes target identification and therapeutic discovery

Machine learning (ML) is ushering in a new era in target identification and therapeutic discovery, with profound implications across various scientific domains. In the realm of proteomics, we have leveraged the power of ML to develop a leading-edge deep learning model called InstaNovo. This model enables high-precision de novo peptide sequencing, eliminating many of the constraints of conventional methods and opening up new opportunities in antibody sequencing, identification of neo-epitopes in cancer, and the exploration of the dark proteome. However, beyond target identification, ML is also proving promising in the rapid discovery and development of therapeutics. Particularly with the rise of generative de novo protein design, design of functional binders entirely in silico has been brought within reach. We have taken advantage of these developments, to design minibinders (small binding proteins primarily comprised of beta sheets and alpha helices) that can neutralise snake venom toxins. Though further validation is needed, these findings hold promise for the rapid development of next-generation therapeutics.