JOVANA
Library Glossary Getting Started Three Levels Fields How it works Mission
Join the mission
All guides

Finding the Genes: QTLs, GWAS, and Polygenic Scores

If hundreds of genes shape a trait, how do we locate them? Tour quantitative trait loci, genome-wide association studies, and the polygenic scores built from them.

From whole genome to suspect regions

If a polygenic trait is spread across many genes, no single Punnett square will find them. Instead geneticists hunt for a quantitative trait locus (QTL): a stretch of the genome — a locus — where genetic differences are statistically linked to differences in the trait. A QTL doesn't pin down one gene; it flags a region that carries one or more variants nudging the trait up or down.

The modern, high-resolution version is the genome-wide association study (GWAS). It scans millions of common single-letter variants — single-nucleotide polymorphisms, or SNPs — across tens of thousands of people, asking at each one: does carrying this variant shift the trait, on average? The answer for any one SNP is usually a tiny effect, which is exactly what we expect when hundreds of variants each contribute a sliver.

GWAS, one SNP at a time (toy numbers):

  SNP at position chr2:18,455,201, alleles G / A
  --------------------------------------------------
  genotype   mean height (cm)   count
     GG          171.0           4200
     GA          171.4           5100
     AA          171.8           2300
  --------------------------------------------------
  Each 'A' adds ~0.4 cm on average -> tiny, but real.

  Repeat across ~1,000,000 SNPs. A handful pass the strict
  genome-wide significance bar (p < 5e-8). Together, many
  small effects add up to a meaningful chunk of variation.
GWAS tests each variant for a small average shift in the trait.

Adding up the effects

Once a GWAS has estimated the effect of each variant, you can add them up for a single person to get a polygenic score: count each trait-raising allele the person carries, weight it by its estimated effect, and sum. The result is a single number estimating that individual's genetic predisposition for the trait relative to the population.

There is an honest gap to name. GWAS often explains less of a trait's variation than heritability estimates suggest it should — the famous “missing heritability.” Some of it hides in rare variants GWAS doesn't catch, some in many ultra-small effects below the significance bar, some in subtle gene–environment interaction. The map of a complex trait is real but still blurry at the edges.