Protein Engineering and 3D Printing
Over two decades ago, a groundbreaking achievement led by David Baker at the University of Washington marked the inception of de novo protein design. While ‘Top7,’ an early creation, initially lacked biological functionality, today, the field has evolved into a powerful tool for tailor-made enzymes and proteins. Neil King, collaborating with Baker’s team, emphasizes the transformative nature of this progress, citing the ability to accomplish tasks that were deemed impossible just a year and a half ago. The success owes much to extensive datasets linking protein sequences to structures and the indispensable role of deep learning in unraveling the hidden grammar of protein architecture.
Deep Learning for Protein Design
The utilization of large language models (LLMs), such as those powering ChatGPT, has paved the way for ‘sequence-based’ strategies in protein design. Noelia Ferruz and her team at the Molecular Biology Institute of Barcelona developed ProtGPT2, an algorithm consistently producing synthetic proteins with stable folding. Meanwhile, ‘structure-based’ approaches, fueled by diffusion models akin to those in image-generating tools like DALL-E, demonstrated notable progress in bespoke protein-design algorithms in 2023. RFdiffusion software and the Chroma tool exemplify the power of these approaches in engineering novel proteins.
Continue reading… “Advancements and Challenges: Seven Key Technological Frontiers Explored”
