Razib Khan's Unsupervised Learning Alex Young: IQ, disease and statistical genomics
Dec 9, 2025
In this captivating discussion, Alex Young, an Assistant Professor at UCLA and a pioneer in statistical genomics, delves into the fascinating world of heritability and genetics. He and Razib explore how advancements in genome-wide association studies (GWAS) have transformed our understanding of traits like intelligence and autism. Young addresses the concept of missing heritability, the complexities surrounding polygenic scores, and the profound implications of embryo genomic prediction at Herasight. Their conversation challenges existing norms while emphasizing the necessity of studying sensitive genetic topics.
AI Snips
Chapters
Transcript
Episode notes
GWAS vs Classical Heritability
- GWAS finds correlations between common genetic variants and traits but often spreads effects across many loci, especially for cognitive traits.
- SNP heritability from GWAS is a different, usually lower, quantity than classical twin heritability and bounds polygenic prediction.
SNP Heritability Limits Prediction
- SNP heritability measures explained variation from common variants available to GWAS and limits polygenic score performance.
- Large GWAS (millions) can make polygenic scores approach SNP heritability, as shown for height.
Family Genomics Reframes Heritability
- Twin studies estimate high heritability but can differ from genomic methods due to confounding like assortative mating.
- Newer family-based genomic methods give more robust heritability estimates that may revise twin-based numbers.