SNP-based Heritability Estimation for Complex Traits

This direction is merged into another direction: missing heritability, SNP-based heritability estimation. Its last version stays readable and citable here.

Version 1 of 121 papers21 conceptsfrozen 2026-10-04

For decades, geneticists faced a paradox known as "missing heritability." Twin studies consistently showed that traits like height, intelligence, and disease susceptibility were highly heritable—often 50% to 80%. Yet, when researchers began scanning the genome for specific markers (Genome-Wide Association Studies, or GWAS), the individual "hits" they found explained only a tiny fraction of that variance, often less than 5%.

The solution, pioneered through methods like Genome-wide Complex Trait Analysis (GCTA), was to stop looking for individual needles in the haystack and instead measure the weight of the entire haystack at once.

The Intuition: Genomic Similarity Among Strangers

Traditional heritability studies rely on known relationships, like twins or siblings. We know exactly how much DNA they share. Dr Maciej Trzaskowski, Jian Yang, and Peter M. Visscher shifted this paradigm by looking at "unrelated" individuals.

Even if two people are not relatives, they still share minute segments of DNA by chance. By analyzing hundreds of thousands of single-nucleotide polymorphisms (SNPs) across the entire genome, researchers can calculate a Genetic Relationship Matrix (GRM). This matrix quantifies exactly how much more (or less) DNA any two people share compared to the population average.

The core logic is elegant: if people who are slightly more genetically similar also happen to be slightly more phenotypically similar, then the trait must be influenced by the aggregate effect of those common SNPs. As Trzaskowski and Robert Plomin demonstrated in DNA evidence for strong genetic stability and increasing heritability of intelligence from age 7 to 12, this method allows us to estimate heritability without needing family data, bypassing the environmental confounding factors that often plague twin studies.

The Mechanism: Restricted Maximum Likelihood (REML)

To turn these tiny correlations into a single heritability estimate, researchers use Restricted Maximum Likelihood (REML). In this framework, the entire genome is treated as a single "random effect" in a linear mixed model.

Instead of testing if SNP A or SNP B causes a trait, the model asks: How much of the total variance in this trait can be explained by the total variance in the genetic similarity matrix?

This approach has been used to uncover the genetic architecture of diverse conditions:

  • Obesity: Clare Llewellyn and colleagues used GCTA to show that common SNPs explain 30% of the variance in childhood BMI, bridging the gap between GWAS hits and twin estimates in Finding the missing heritability in pediatric obesity.
  • Brain Structure: Christopher Bryant and Joseph G. Ibrahim applied these models to MRI data, finding that while overall head size is 85% heritable, specific sub-regions like the caudate nucleus have their own distinct genetic "blueprints" (Mapping the Genetic Variation of Regional Brain Volumes).
  • Disease Risk: For conditions like Benign Prostatic Hyperplasia, Jacklyn N. Hellwege used REML to show that 60% of the variation is genetic, even when no single gene reached statistical significance (Heritability and genome-wide association study of benign prostatic hyperplasia).

Why It Works: Capturing Polygenicity

The reason GWAS often "fails" to find genes is that most complex traits are polygenic—influenced by thousands of variants, each with an effect so small it is indistinguishable from noise. SNP-based heritability works because it doesn't try to filter the noise. It sums the signal across the whole genome.

This aggregate view has revealed that many traits are not just polygenic, but pleiotropic—the same genes influence multiple traits. Trzaskowski and Oliver S. P. Davis showed that the DNA markers for general intelligence are largely the same as those for reading and math, providing molecular evidence for "general cognitive ability" (DNA Evidence for Strong Genome-Wide Pleiotropy of Cognitive and Learning Abilities).

Beyond Static Estimates

Recent work emphasizes that SNP-based heritability is not a fixed biological constant but a dynamic property of populations. Felix C. Tropf and colleagues demonstrated through a mega-analysis of 31,000 individuals that the genetic influence on fertility shifts depending on the historical and geographical context, pointing to powerful gene-environment interactions (Mega-analysis... uncovers strong gene-environment interaction for human fertility).

Furthermore, as datasets grow to biobank scales, new computational tools like the Sparse Marginal Epistasis Test (implemented in the smer package by Julian Stamp) are allowing researchers to move beyond simple additive effects to capture the "missing heritability" hidden in complex gene-gene interactions (smer: Sparse Marginal Epistasis Test).

By shifting the focus from individual genes to the collective genomic background, these methods have transformed our understanding of how DNA shapes everything from the structure of our transmodal cortex to our risk of drug-induced liver injury.

Papers in this version

History & citationVersion 1 · 2026-10-04

Merged into another direction · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_snp_based_heritability_estimation_for_co_e4f8dedd,
  title = {SNP-based Heritability Estimation for Complex Traits},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/snp-based-heritability-estimation-for-complex-traits/rdr_e4f8dedd609b4504b168b324b0d1bf9b},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_snp_based_heritability_estimation_for_co_e4f8dedd_v1,
  title = {SNP-based Heritability Estimation for Complex Traits},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/snp-based-heritability-estimation-for-complex-traits/rdr_e4f8dedd609b4504b168b324b0d1bf9b/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}