GWAS (Genome-Wide Association Study)
A statistical approach that scans hundreds of thousands of genetic variants across thousands of genomes to identify variants associated with a disease or trait — the foundation of modern human genomics and polygenic risk research.
What it means
A genome-wide association study (GWAS) is a statistical analysis that tests hundreds of thousands to millions of single nucleotide polymorphisms (SNPs) — single-letter DNA variants — across a large group of individuals to identify variants statistically associated with a disease, trait, or other outcome.
The basic design: collect DNA and phenotype data from many thousands of people (cases with the disease + controls without it, or a continuous trait in a population cohort). At each SNP position in the genome, test whether people who carry the variant are more or less likely to have the trait. Apply a stringent significance threshold (p < 5×10⁻⁸, accounting for ~1 million simultaneous tests) to avoid false positives from multiple testing.
What GWAS finds: Common variants (minor allele frequency > 1–5%) with modest effects on complex traits. Most GWAS hits have odds ratios of 1.05–1.2 — small effects, but detectable because the variants are common and sample sizes are large (100,000+ in modern GWAS).
What GWAS does not find: Rare variants with large effects (those require different study designs like whole-exome sequencing); causal mechanisms (an associated variant may be in linkage disequilibrium with the actual causal variant); environmental interactions.
The missing heritability problem
For most complex traits (height, BMI, schizophrenia, educational attainment), the sum of all known GWAS hits explains far less variance than twin studies suggest is heritable. This “missing heritability” is attributed to many common variants of very small effect, rare variants not captured by standard arrays, and gene-environment interactions.
Polygenic risk scores (PRS) aggregate thousands of small-effect GWAS variants into a single predictive score and partially recover this missing variance.
AI and GWAS
AI tools for variant interpretation (AlphaMissense, Evo 2) use GWAS summary statistics as training signal — variants with larger GWAS effects in relevant traits are more likely to be functionally important.
Deep learning for GWAS: Neural network approaches like DeepSEA and Enformer predict regulatory variant effects from DNA sequence, complementing GWAS by predicting functional impact rather than just statistical association.
LLMs for GWAS literature: When using AI assistants to help with GWAS results, be cautious about claims regarding specific GWAS-identified genes and their functions. LLMs frequently confuse GWAS associations (statistical, in non-coding regions) with gene function knowledge (mechanistic, based on experimental evidence).