New effort launches at Broad to accelerate translation of genetics to therapies

The promise of human genetics has been to reveal which genes are linked to human diseases, and to use that knowledge to create new therapies and improve human health. While researchers have discovered genetic factors contributing to virtually all diseases, progress at figuring out how those genes increase disease risk, and using that insight to develop treatments, has been slow.

One reason is that biologists typically study disease-linked genetic variants one-at-a-time. Even when these studies can be scaled up using high-throughput technologies, it can still take years to say with certainty "variant A affects gene B and perturbs pathway C, and by targeting protein D we can treat this disease."

Broad Institute scientists led by institute member Mark Daly have launched an initiative aimed at solving this challenge. With initial support from The Klarman Family Foundation, the project, called GaMBiT (Genes, Mechanisms, Biomarkers and Therapeutics), brings together scientists, key technologies, and other resources to accelerate the translation of genetic discoveries into therapeutics. The goal is to build a scalable and systematic research pipeline that any scientist can access to discover disease mechanisms, identify biomarkers of disease biology, and pursue promising drug targets.

To learn more about GaMBiT and its goals, we spoke to Daly, who is co-director of Broad's Program in Medical and Population Genetics and chief of the Analytical and Translational Genetics Unit at Massachusetts General Hospital.

For two decades, it's been said that genetics will revolutionize medicine. Has that been the case?

Mark Daly: Not as much as we would have hoped. At this point, genetic discovery is straightforward, and can be done successfully for almost any disease. But the interpretation of genetic variants associated with disease remains challenging, especially for common and chronic diseases like schizophrenia, Parkinson’s, or inflammatory bowel disease, where hundreds of variants each make small contributions to risk.

This gap between discovery and interpretation has prevented us from realizing the original promise of human genetics, and left us largely stuck at the starting gate.

Why hasn't that promise been fulfilled?

MD: The root of the challenge is uncovering which molecular processes disease-associated variants perturb, and how those processes in turn impact disease mechanisms — most of which we have not yet discovered! Keep in mind that in diseases with hundreds of contributing factors, finding and understanding disease mechanism requires looking at many variants and revealing patterns in the processes they affect, not simply each variant in a vacuum.

What’s also been challenging is integrating what we learn about causal mechanisms with proteomic, epigenomic, immune, or metabolomic data directly from humans with and without disease, or at different stages of disease. Those data provide measurable readouts of disease processes, as well as of environmental exposures that influence disease risk and human biology in general. And this is crucial. It's highly unlikely that someone would take a new mechanism, develop a therapy that targets it, and design a clinical trial unless there are measurable human biomarkers that recapitulate that mechanism and tell you whether you've pushed the system in the right direction.

Additionally, we need to be able to systematically integrate genetic variation with protein science, chemical biology, and other fields. That gives us the ability to understand biochemically what mutations may actually be doing, which will in turn help to identify opportunities to develop targeted small molecules, antibodies, or other potential approaches that will have a therapeutic impact.

How will GaMBiT address these challenges?

MD: It provides an umbrella. All of the capabilities and expertise needed to establish mechanisms, biomarkers, and targets for any disease are here at Broad. GaMBiT creates a framework for integrating, scaling, and focusing these capabilities on specific diseases in a comprehensive manner. We picture a scenario where an investigator with a disease project can dock directly into established discovery and development programs, rather than working one step at a time with separate labs and institutions.

So starting with biospecimens, clinical data, and genetic maps for a particular disease, what could be delivered at the end would be experimental models of disease processes, readable biomarkers of mechanism, relevant targets, and, ultimately, compounds, antibodies, or other potentially therapeutic molecules.

Importantly, we are designing this such that every component — editing massive numbers of variants into cells and reading out mechanisms, -omics studies, etc — will feed back into a growing, queryable knowledge base. So for every new disease we study and every experiment performed, we learn more about the relationship between genetic variants and mechanisms, biomarkers and disease, which will over time make the whole process of genetic translation faster and more robust.

In the long run, I think this will be one of the greatest benefits of GaMBiT. All of the activities and knowledge that are now siloed across many labs with distinct technical expertise or disease interests would become part of a learning ecosystem and refine the process going forward.

Why is now the right time for something like GaMBiT?

MD: The technologies we need have really matured in the last few years. Even five years ago, many of these capabilities weren't there. And furthermore, before the advent of LLMs and agentic AI, we didn't have any real mechanism for thinking about how you would integrate across all of these different data types at scale. Now it's totally credible to create huge, multi-dimensional datasets and wring more insight out of them than has ever been possible.

As GaMBiT gets off the ground, what are your hopes for it?

MD: My hope is that we can move beyond the current state where more than 95 times out of 100, our genetic discoveries don't progress to concrete and impactful knowledge. That we can much more systematically identify and attach disease mechanisms to variants and attach biological readouts to those mechanisms. With that, we start to think more holistically about disease and solutions, and not just about individual genetic contributors.

What are the next steps for the initiative?

MD: We are truly grateful for the opportunity this support creates as an institution, and are hitting the ground running. We have already begun to build the computational infrastructure that will undergird the effort, assemble a team to manage and oversee the initiative, and actively discuss what our first flagship disease efforts may be. And we're actively looking to bring people into the fold! Anyone who wants to be part of this should reach out to me.