Precision medicine is useful for gaining a deeper understanding of patient biology, identifying subpopulations, and matching the right therapies to the right patients. David Gnutt is exploring how precision medicine can uncover new targets for heart failure. He stressed the urgent need for this since cardiovascular diseases are still one of the leading causes of death across the globe.
To address the burden of cardiovascular disease, it is crucial to understand patient biology in more detail. On one hand, this could involve encompassing the genetic understanding of patients as well as the genetic drivers of disease.
In some cases, scientists focus on causal therapies in rare diseases, like dilated cardiomyopathy. Whereas in other cases, they try to understand broader patients in more detail, for instance, who are the responders and fast progressors, because this will make it easier to match the right patient to the right medication.
Gnutt advocated for using multiomics to understand patient biology. So, the team integrated multiple data sources, including omics data and patient records, to build knowledge graphs of patient biology at an unprecedented scale. This ultimately predicts and prioritizes new drug targets for cardiovascular diseases.
Researchers have integrated foundational data into a general model using transfer learning to make in silico predictions about heart disease. Based on clustering patterns, this model distinguishes between different heart conditions like non-failing, dilated cardiomyopathy, and hypertrophic hearts. By simulating genetic changes in silico, the model can predict targets that might revert a diseased heart back to a healthy state.
Each prediction round generates 300–400 potential targets, which are then evaluated using cell painting, which is a scalable imaging technique. Next, the researchers stained iPSC-derived cardiomyocytes from both healthy and diseased lines and analysed them using custom cell profiler pipelines to extract hundreds of features and unbiasedly cluster phenotypes.
This method is highly scalable and allows for high-throughput screening of perturbations, some of which visibly disrupt sarcomere structure, critical for heart contraction. This highlights the approach’s applicability in functional validation.
There are ongoing challenges, such as druggability and the safety of targets. Furthermore, it is essential to gather complementary data layers to expand drug discovery entry points and accelerate decision-making.
