CRISPR image from Ernesto del Aguila III, National Human Genome Research Institute, NIH
The potential of Artificial Intelligence (AI) to accelerate medical discovery is immense, particularly within the dynamic and mutable area of biology. Unlike the more predictable fields of chemistry and physics, biology presents unique challenges due to its complexity and constant adaptation, areas where AI can play a transformative role but has been limited to date due to this dynamism.
Historically, the application of AI has concentrated in fields like chemistry and physics because, while complex, their predictability has enabled early successes which have attracted significant investment. The mutable nature of biological systems made them nearly impossible to model effectively, leading scientists and entrepreneurs to focus on the so-called “low-hanging fruit” within more stable disciplines. That’s not to say the use of AI hasn’t been fruitful. AlphaFold, DeepMind’s AI program is used to help identify the perfect binding spots on cancer-causing proteins, allowing us to rationally design drugs that better target them. But we need many more AlphaFolds to take on massive medical challenges. I believe that using AI more in biology can enable us, as scientists, policymakers, investors and entrepreneurs, to do much more and take bigger swings.
For example, cancer metabolism, particularly the expansive terrain of the kinome and metabolic genome, is an area of almost infinite variables that, without AI, is almost impenetrable. AI holds the key to moving beyond the well-characterized kinome into the vast, largely unexplored metabolic genome. This part of the genome plays a crucial role in cancer’s energy and nutrient utilization. Historically, cancer research has focused extensively on the kinome—the set of protein kinases within the human genome that are considered druggable targets. However, about 50% of the druggable genome, particularly those parts related to metabolism, offers huge potential to reveal completely new treatment avenues.
Cancer and metabolism alone involve approximately 3,000 drugs, 200 metabolites, and 10,000 genes, resulting in over six billion possible interactions. AI is the only way to manage this scale and complexity by modeling these interactions and offering opportunities to identify novel and promising therapeutic targets. AI can help reduce the likely combinations down from the hundreds of thousands to the tens and help pinpoint these targets. CRISPR technology can then be used to experimentally validate these findings with precision, feeding empirical data back into AI models to refine predictions and hypotheses further. The most promising results from these CRISPR experiments can be further culled to the most viable few, where human researchers can undertake detailed investigations. By narrowing down billions of potential interactions to a manageable few, researchers can focus their efforts on the most promising avenues for therapeutic development.
Beyond cancer metabolism, AI’s capabilities also extend to neurological, endocrine, immunology and inflammatory diseases. In these areas, many researchers focus on a limited set of known targets. AI enables them to cast a much wider net, examining thousands of potential therapeutic agents, genes, and nutrients that can influence the human microenvironment. This broad examination could redefine treatment strategies for a range of diseases, paving the way for innovations that were previously constrained by the limits of traditional research methodologies.
Embracing AI in biology means tackling some of the most complex and variable aspects of science for the very first time. By focusing AI’s power on the untapped potential of the metabolic genome in cancer research, for example, we are not just making incremental improvements but are positioned to make quantum leaps in understanding, managing and treating cancer. This is key to unlocking novel strategies that could serve as the fourth pillar of cancer treatment, alongside surgery, chemotherapy, and radiation. This innovative approach could fundamentally transform cancer care, providing personalized, effective treatments based on the metabolic profiling of individual tumors and a patient’s likely reactions to multiple therapeutic approaches.
By doing so, we are setting the stage for discoveries that could fundamentally alter our approach to health and disease, making biology the next great frontier for AI.
Anand Parikh, JD is CEO and co-founder of Faeth Therapeutics.
Filed Under: Data science, Drug Discovery and Development, Omics/sequencing, Oncology
