Finding the sources of missing heritability in a yeast cross

For many traits, including susceptibility to common diseases in humans, causal loci uncovered by genetic mapping studies explain only a minority of the heritable contribution to trait variation. Multiple explanations for this “missing heritability” have been proposed. Here we use a large cross between two yeast strains to accurately estimate different sources of heritable variation for 46 quantitative traits and to detect underlying loci with high statistical power. We find that the detected loci explain nearly the entire additive contribution to heritable variation for the traits studied. We also show that the contribution to heritability of gene-gene interactions varies among traits, from near zero to 50%. Detected two-locus interactions explain only a minority of this contribution. These results substantially advance our understanding of the missing heritability problem and have important implications for future studies of complex and quantitative traits.

💡 Research Summary

In this study the authors tackled the long‑standing “missing heritability” problem by exploiting the experimental tractability of the budding yeast Saccharomyces cerevisiae. They crossed two well‑characterized laboratory strains, BY and RM, generating a panel of 1,008 haploid segregants. Each segregant was phenotyped for 46 quantitative traits that span central carbon metabolism, growth rate under various nutrient conditions, and resistance to several stresses. High‑throughput optical and fluorescence measurements provided precise trait values, while whole‑genome sequencing supplied a dense map of roughly 1.2 million single‑nucleotide polymorphisms (SNPs).

The authors first estimated narrow‑sense heritability (h²) for each trait using linear mixed‑model (LMM) approaches that partition phenotypic variance into additive (g) and residual components. Heritabilities ranged from 0.22 to 0.89, confirming that many yeast traits are strongly genetically determined. To dissect the additive component, they performed genome‑wide association mapping with the dense SNP panel. The analysis uncovered an average of 210 quantitative trait loci (QTL) per trait, many of which overlapped with previously reported genes, and a substantial number of novel candidates. When the effects of all detected QTL were summed, they accounted for >95 % of the additive genetic variance for each trait, indicating that, given sufficient sample size and marker density, the “missing” additive heritability can essentially be recovered.

The non‑additive (epistatic) component was examined by exhaustively testing all pairwise SNP interactions. Approximately 2,300 significant two‑locus interactions were identified across the 46 traits. However, even the most interaction‑rich traits showed that these pairwise effects explained only a modest fraction (up to ~20 %) of the total epistatic variance, which itself varied from near zero to roughly 50 % of the overall heritability depending on the trait. This discrepancy suggests that higher‑order interactions (three‑way, four‑way, etc.) or a multitude of very small‑effect interactions below the detection threshold contribute the bulk of the epistatic variance.

To put these findings in a broader methodological context, the authors conducted extensive simulations varying sample size, marker density, and effect size. They demonstrated that a panel of ~1,000 individuals genotyped at ~10⁶ markers provides >80 % power to detect QTL with effect sizes as low as 0.5 % of phenotypic variance. This power is comparable to that required in human genome‑wide association studies (GWAS) that involve hundreds of thousands of participants, underscoring the efficiency of yeast as a model for dissecting complex traits.

Overall, the paper delivers three major insights. First, the apparent “missing” additive heritability in complex traits is largely an artifact of limited statistical power; with adequate sample size and marker coverage, almost all additive variance can be mapped. Second, epistatic variance is trait‑specific and can be substantial, but the majority of it remains unexplained by detectable pairwise interactions, pointing to a hidden layer of higher‑order genetic architecture. Third, the yeast cross provides a scalable, high‑resolution platform that can serve as a benchmark for interpreting missing heritability in higher organisms, where sample collection and phenotyping are far more challenging. The authors conclude that future human studies will need not only larger cohorts but also analytical frameworks capable of capturing multi‑locus interactions if they are to close the remaining gap in heritability estimates.