Whole genome sequencing reveals how much human heritability we can finally explain

By Vijay Kumar MalesuReviewed by Susha Cheriyedath, M.Sc.Nov 13 2025

Sequencing nearly half a million genomes, researchers show that most additive genetic influences on height, lipids, and other complex traits are now directly measurable, while pinpointing ultra-rare and structural variants as key suspects behind the heritability that remains missing.

Study: Estimation and mapping of the missing heritability of human phenotypes. Image Credit: Buntan / Shutterstock

In a recent study published in the journal Nature, a group of researchers quantified the contribution of rare and common coding and non-coding variants, as measured by whole-genome sequencing (WGS), to heritability across diverse human traits, and assessed the extent to which pedigree-based narrow-sense heritability they collectively explain.

Limits of GWAS and WES for Genetic Variation

Most human traits show some heritability, yet the deoxyribonucleic acid (DNA) changes behind this family resemblance remain elusive. Twin and family studies suggest strong additive genetic effects, but genome-wide association studies (GWAS) using common single-nucleotide polymorphisms (SNPs) explain only a modest share of variation. Whole-exome sequencing (WES) highlights rare coding variants but covers only under 3% of the genome. WGS can survey rare variation across coding and non-coding regions; however, their contribution to trait differences remains unclear. Further research is needed to measure how much missing heritability these rare variants explain.

Large Scale UK Biobank Sequencing Dataset

The investigators used WGS data from 490,542 United Kingdom Biobank (UKB) participants and focused on 347,630 unrelated individuals of European ancestry. They analyzed 40.6 million autosomal SNPs and insertion-deletion (indel) variants with a minor allele frequency (MAF) above 0.01%. For 41 complex traits and diseases with measurable pedigree-based heritability, they estimated WGS-based heritability using genetic restricted maximum likelihood (GREML) with linkage disequilibrium (LD) and minor allele frequency stratification, implemented in the Mixed Model and Partitioning Heritability (MPH) software. They decomposed total heritability into contributions from rare variants with MAF below 1% and common variants with higher frequency. Further, they partitioned these components into coding and non-coding genomic annotations.

Pedigree Heritability and GWAS Mapping Strategy

Pedigree-based narrow-sense heritability was estimated from 171,446 related pairs, allowing direct comparison with sequencing-based estimates. The team also conducted GWAS in 452,618 individuals to map independent common and rare variant associations, correcting for the winner’s curse. Analyses adjusted for age, sex, genetic principal components, and birthplace clusters, and focused inference on 34 phenotypes with significant WGS and rare variant heritability. This work spanned both quantitative traits and common diseases. Notably, for behavioral traits such as educational attainment and fluid intelligence, the study required additional correction for assortative mating and geographic stratification to avoid inflated heritability estimates. This nuance underscores the complexity of dissecting genetic contributions for specific phenotypes.

Rare Variant and Coding Region Heritability Patterns

Across 34 traits with rare variant signals, WGS-based heritability estimates ranged from 8% for the number of children to 71% for height, averaging 28%. Rare variants with MAF below 1% accounted for about 20% of pedigree-based heritability, while common variants contributed 68% of pedigree-based heritability, leaving approximately 12% still missing. Thus, within the 88% of heritability captured by WGS, rare variants account for approximately 23%, and common variants account for 77%. Coding regions represented less than 1% of all analyzed variants yet contributed roughly 17% of total WGS heritability and about 20% of rare variant heritability, implying a 26- to 36-fold enrichment for functional coding changes. Non-coding variants, although individually weaker, collectively explained most of the rare variant contribution.

Trait Specific Differences in WGS Heritability

When sequencing-based heritability was compared with pedigree-based estimates from 171,446 relative pairs, WGS captured 88% of narrow-sense heritability on average. For 15 highly powered traits, there was no significant difference between sequencing-based and pedigree-based estimates, suggesting that most of their additive genetic variation is now accounted for by observed variants. In contrast, traits such as number of children and telomere length retained substantial heritability, hinting at roles for ultra-rare variants, structural variation, or non-additive genetic effects. The study also cautioned that including ultra-rare variants (MAF less than 0.01%) produced negative heritability estimates for some traits, a classic sign of model misspecification, indicating that current methods are not yet reliable for this variant class.

Rare Variant Associations and Structural Variation

The explained heritability ratio, defined as the ratio of WGS-based to pedigree-based heritability, averaged 0.88 but varied across traits, illustrating that genetic architecture and statistical power shape how complete current maps have become. GWAS using the same genomes identified more than 12,000 independent associations, including 11,243 common variant hits and 886 rare variant hits across 34 traits. Each rare association explained more phenotypic variance than each common association. Rare signals were informative for lipid traits, where low-density lipoprotein (LDL) cholesterol and high-density lipoprotein (HDL) cholesterol had rare variant associations that, together, explained more than 33% of their rare variant heritability. Alkaline phosphatase (ALK) was the only non-lipid trait showing similarly high explanatory power. Many of these rare associations lie in or near loci already flagged by common variant signals, showing colocalization.

Structural Haplotypes and Missing Sequence

In several regions, clusters of rare and common associations coincided with structural variants, reinforcing the idea that complex haplotypes underlie some of the strongest genetic effects. The analysis also revealed that previous work suggests the X chromosome contributes less than 3% to heritability, and that the hg38 genome build used misses approximately 8% of the DNA sequence, both factors that account for a small fraction of the still-missing heritability.

Clinical Relevance of WGS Heritability Mapping

Overall, the results indicate that a measurable portion of the previously missing heritability is already mappable with fewer than 500,000 fully sequenced genomes, particularly for well defined traits such as blood lipids and liver enzymes. For patients and clinicians, these findings reasonably suggest more precise genetic risk scores, earlier detection of high-risk profiles, and better-targeted prevention strategies worldwide in practice, although clinical translation was not directly examined in this study.

Remaining Gaps and Future Genomic Research

This study demonstrates that WGS can recover most of the narrow-sense heritability previously inferred from family-based designs for many human traits. By dividing heritability into rare and common as well as coding and non-coding components, it narrows the space left for unknown genetic factors and clarifies where missing heritability still remains. For families, clinicians, and policymakers, the findings suggest that polygenic scores incorporating rare variants could improve risk prediction, especially for lipid-related traits. Closing the remaining gap will require larger, ancestrally diverse global cohorts and improved tools for ultra-rare and complex structural variation.