Finding a Match Between Genomics and Machine Learning

Genomics modeling is notoriously challenging due to the extremely high-dimensional nature of the data. With sample sizes often being small and results requiring high interpretability and reliability, classical statistical methods remain the gold standard. But can we leverage modern computational approaches to push the boundaries of genomics research?

Recently, I had a great discussion with Karin Hrovatin, Alan, and Yasha exploring this exact puzzle. We discussed topics ranging from increasing GWAS power using deep learning priors, to my work on efficiently detecting marginal epistasis, and leveraging evolutionary protein embeddings for phylogenetic tree reconstruction.

Karin wrote a fantastic and comprehensive blog post summarizing our discussion. Curious about the details? Dive into her article here: Finding a match between genomics and machine learning | Karin Hrovatin