How HPC and generative AI will accelerate drug discovery and precision medicine

Two decades ago, humanity first sequenced its own genetic code, the start of an unprecedented look at the roles and interactions of amino acids, proteins, and other building blocks of life. From both a scientific and healthcare basis, we’ve arrived in a new world of knowledge and treatment, based on discerning the code of life. It is a profound information problem, in which research understanding and therapeutic breakthroughs only come after years of painstaking and often tedious work. For example, many new drugs are now coming to market after 10-15 years of work and hundreds of millions of dollars later. 

Artificial intelligence, high performance computing, and AlphaFold are among the new tools for advanced computations and drug discovery. Many of our customers are leading the way, including Bayer, Servier, and Schrödinger. And, earlier today, we announced the Target and Lead Identification Suite, an AI tool for life science researchers to better identify the function of amino acids and predict the structure of proteins; and the Multiomics Suite, for discovering and interpreting genomic data, and designing personalized genomic treatments. Pfizer, Cerevel, and CSIRO, Australia’s national science agency, are using the Target and Lead Identification Suite, and Colossal Biosciences is adopting the Mutiomics Suite. Organizations around the world can take advantage of these new solutions with our global delivery partners that reach every market, including EPAM Systems, Inc, Form Bio, Max Kelsen, Omnigen, Quantiphi, and others.

Generative AI advancing life sciences

Looking ahead, we foresee new ways to augment existing tools, including the Target and Lead Identification Suite and the Multiomics Suite, and build entirely new ways to advance the field of life sciences, using the growing discipline of generative AI. In fact, we see it occupying a uniquely powerful place in the toolset.  

Generative AI, which moves AI from analysis to creation of new information, is generally based on large models (LMs), which look at the billions of interactions that create communications, whether that’s words in speech or pixels in images. Once trained on an LM, generative AI is capable of digesting and summarizing information, and creating new information, all based on sophisticated probabilistic analysis and a lot of specialized computing.

One of the LMs we expect to help advance life sciences is Med-PaLM 2, an LM fine-tuned for the medical domain and able to more accurately and safely answer medical questions. We recently announced we’re providing limited access to Med-PaLM 2 to a select group of Google Cloud customers, and we expect valuable use cases to come from their feedback. 

While we’ll have some innovations like Med-PaLM 2 that are tuned for healthcare and life sciences, we also have products that are relevant across industries. In March, we announced several generative AI capabilities coming to Google Cloud, including Generative AI support in Vertex AI to build and deploy generative AI applications at scale and Gen App Builder to help organizations build their own AI-powered chat interfaces and digital assistants in minutes or hours by connecting conversational AI flows with out-of-the-box search experiences and foundation models. We announced even more generative AI services for Google Cloud customers in early May at Google I/O 2023. 

What’s more, when Google Cloud brings new AI advances to our products, our commitment is two-fold: to not only deliver transformative capabilities, but also ensure our technologies include proper protections for our organizations, their users, and society. To this end, our AI Principles, established in 2017, form a living constitution that guides our approach to building advanced technologies, conducting research, and drafting our product development policies. 

The future of complexity

There is much, much more to understand about the code of life. Soon after the human genome was decoded, we discovered the complexities of the underlying proteins, and the many characteristics and interactions underlying that. Today’s announcements are groundbreaking, but many puzzles lie ahead, and there’s much understanding to be gained. 

It’s clear that nature won’t be running out of complexity any time soon, so there’s a boundless amount left to learn. We couldn’t be more committed and excited to deliver new tools for new and better understanding and treatments.