EP 238: Uncovering epistatic interactions in complex disease with machine learning with Bin Yu of UC Berkeley

This week on The Genetics Podcast, Patrick is joined by Dr. Bin Yu, CDSS Chancellor’s Distinguished Professor at UC Berkeley. They discuss how different statistical approaches, from linear models to random forests, can be used to study complex genetic traits, recent findings on epistasis in cardiomyopathy, and how improving robustness and reproducibility can lead to more reliable scientific conclusions.

Show Notes

0:00 Intro to The Genetics Podcast

01:00 Welcome to Bin

01:47 Linear models as the foundation of genetic analysis

05:34 Using random forests and stability to identify gene–gene interactions beyond linear models

11:05 How iterative feature weighting in random forests improves detection of gene interactions

13:10 Using GWAS to prioritize features in high-dimensional genetic data

15:06 Applying stable interaction models to hypertrophic cardiomyopathy in UK Biobank

20:47 Biological insights from gene–gene interactions in cardiomyopathy and evidence for indirect epistasis

23:25 Scaling discovery of epistatic interactions with better data and integrated experimental validation

27:21 The predictability, computability, and stability (PCS) framework for data science

30:06 How Bin’s early life during the Chinese Cultural Revolution shaped her 

32:54 Balancing AI-driven productivity with human reasoning and scientific thinking

35:23 Developing the ability to read people through observation, listening, and real-world interaction

38:03 Closing remarks

Find out more:

• Epistasis in cardiac hypertrophy study

• https://vdsbook.com/