📄 evolutionary biology

Rare variants drive high variance in human ancestral fitness at mutation-selection-drift balance

By correcting empirical estimates of deleterious mutation rates to include slightly deleterious variants, this study demonstrates that rare and ultra-rare mutations drive substantial variance in human ancestral fitness, potentially explaining the "missing heritability" in complex diseases and highlighting the need to assess unique mutations rather than relying solely on polygenic risk scores.

Original authors: Hernandez, U., Mawass, W., Matheson, J., Masel, J.

Published 2026-07-31

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Hernandez, U., Mawass, W., Matheson, J., Masel, J.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The Invisible Backpacks We All Carry

Imagine your body is a high-performance race car. For decades, scientists have been obsessed with the big, obvious dents in the paint—the broken parts that clearly cause the car to sputter. These are the major genetic diseases we know about, like cystic fibrosis or Huntington's. But there's a quieter, more mysterious problem: the thousands of tiny, almost invisible scratches on the bumper. Individually, a scratch doesn't stop the car. But if you have a few hundred of them, the car might run slower, overheat, or break down sooner than a pristine model.

In the world of genetics, these "scratches" are called deleterious mutations. They are tiny errors in our DNA that slightly weaken our health or ability to survive. For a long time, scientists wondered: Do these tiny errors pile up differently in every person? Is one person carrying a "heavy" backpack of these errors while another has a "light" one? This question matters because if the backpacks are very different sizes, it could explain why some people get sick easily while others stay healthy, even when they look genetically similar. It also helps explain a mystery in medicine called "missing heritability"—the fact that we can't fully predict disease risk just by looking at the common genetic markers we usually check.

The Great Genetic Backpack Race

In this study, the authors decided to do some serious math to figure out just how heavy these invisible backpacks are for humans. They didn't just look at the big, obvious errors; they built a sophisticated model to count the tiny, rare ones that most people miss. Think of it like trying to weigh a backpack by counting every single grain of sand inside it, rather than just looking at the big rocks.

The researchers used a concept called mutation-selection-drift balance. Imagine a factory (our bodies) that keeps making tiny mistakes (mutations) in its blueprints. Nature acts as a quality control inspector (selection), trying to throw away the bad blueprints. But sometimes, the inspector gets distracted or the factory is too small, and a bad blueprint slips through by chance (drift). The authors asked: In a human population, how much do these "slipped-through" bad blueprints vary from person to person?

Their calculations suggest that the variation is huge. They found that two random humans likely differ in their "ancestral fitness"—how well they would have survived and reproduced in the wild, long ago—by 17% to 33%. If you consider a wider range of possibilities for how bad these mutations are, that difference could swing anywhere from 6% to 49%. To put that in perspective, if one person is a "perfect" runner, the next person might be running with a backpack that makes them 30% slower, purely because of the random accumulation of these tiny genetic errors.

The Power of the Rare

Here is the twist that makes the story so interesting: The authors discovered that the heavy backpacks aren't made of the common, well-known mutations. Instead, the weight comes almost entirely from rare variants—mutations that are so rare they appear in less than 1% of the population, and often in less than 0.01%.

Imagine a library where most books have a few typos. You'd expect the worst books to be the ones everyone reads. But this paper suggests the real troublemakers are the books that only one or two people have ever checked out. These ultra-rare mutations are often the ones with the biggest impact on fitness. The study shows that these rare, unique errors are responsible for about 99% of the variation in fitness when the mutations are strong, and still a massive 83% even when they are weaker.

This finding is a big deal for medicine. Right now, doctors often use "Polygenic Risk Scores" to predict disease, which mostly look at the common mutations (the books everyone has read). The authors suggest that this approach is missing the heavy lifters. To truly find the people at highest risk for disease, we need to look at the unique, ultra-rare mutations in a person's entire genome, not just the common ones.

What This Means for the Future

The paper doesn't claim to have solved the mystery of disease, nor does it say we can now perfectly predict who will get sick. Instead, it suggests that the "missing heritability" we've been looking for might be hiding in these rare, unique genetic quirks. It also clarifies that having a high load of these mutations doesn't mean a species is doomed; the math shows that even with these heavy backpacks, the "best" individuals in a population can still have enough energy to reproduce and keep the species going.

Ultimately, this research paints a picture of human genetic diversity that is far more chaotic and varied than we thought. We aren't just carrying a few common bad genes; we are each carrying a unique, heavy, and mostly invisible collection of rare errors that shape our health in ways we are only just beginning to understand.

Problem
A central question in human genetics is whether variation in the genetic load of unconditionally deleterious mutations contributes substantially to the variability in human disease and fitness. While polygenic risk scores (PRS) based on common variants are standard, they often fail to capture the "missing heritability" observed in pedigree studies. Previous models of mutation load often focused on "lag load" (the difference between mean population fitness and a hypothetical mutation-free individual), leading to the "mutation load paradox" of absurdly high fitness costs. Furthermore, standard estimates of the deleterious mutation rate ( $U$ ) derived from sequence constraint ( $U_{obs}$ ) likely underestimate the true rate because they fail to account for weakly deleterious mutations that fix at neutral rates. Additionally, many models ignore the effects of genetic drift on the variance of load, particularly the probability that sites are monomorphic, which is critical for human genomes where new mutations are rare relative to effective population size.

Methodology
The authors develop a mathematical framework to solve for mutation-selection-drift balance (MSD) rather than simple mutation-selection balance.

Revised $U$ Estimation: They correct the observed deleterious mutation rate ( $U_{obs}$ ) to estimate the total human deleterious mutation rate ( $U$ ). They account for the fact that $U_{obs}$ misses weakly deleterious mutations that fix. Using a Distribution of Fitness Effects (DFE) inferred from human non-synonymous mutations (specifically the ESP sample) and ultra-conserved non-coding regions, they calculate the probability that selection prevents fixation. This yields a correction factor where $U \approx 1.34 U_{obs}$ to $1.68 U_{obs}$ depending on the DFE parameters.
Modeling Variance: They model the variance in log relative fitness ( $\sigma_r^2$ ) across the genome. Unlike previous diffusion approximations that assume sites are always segregating, they incorporate the probability of monomorphism using queueing theory (Poisson process for mutation arrival and mean segregating time).
Parameters: The model utilizes a realistic human per-diploid-site mutation rate ( $\mu = 1.515 \times 10^{-8}$ ), an effective population size ( $N_e = 11,300$ ), and a gamma-distributed DFE with varying mean selection coefficients ( $\bar{s}$ ) ranging from $0.00064$ to $0.01$ and a coefficient of variation (CV) of 2.43.
Metrics: Instead of lag load, they calculate the fitness advantage of a 75th-percentile individual over a 25th-percentile individual (Q75-Q25 advantage) and the absolute fitness of the best individual ( $W_b$ ) to ensure biological plausibility.

Key Contributions

Correction of Mutation Rate: The study demonstrates that the total human deleterious mutation rate is significantly higher than constraint-based estimates, estimating $U > 3.8$ (with an upper bound near 13), largely due to the inclusion of weakly deleterious mutations.
Quantification of Load Variance: The authors provide a rigorous calculation of the variance in ancestral human fitness attributable to unconditionally deleterious mutations under MSD balance.
Role of Rare Variants: The model partitions the variance in load by allele frequency, demonstrating that rare variants (frequency < 1%) and ultra-rare variants (frequency < 0.01%) are the primary drivers of fitness variance.
Drift Effects: The paper clarifies that while drift can increase the frequency of deleterious alleles, it also increases allele loss (monomorphism), which reduces the overall variance in fitness compared to deterministic models.

Results

Fitness Variance: Two typical humans differ in ancestral fitness by 17–33% given uncertainty in $U$ , or by 6–49% when considering a broad range of alternative DFEs (including lower bounds for $\bar{s}$ ).
Source of Variance: Rare variants with frequencies below 1% explain approximately 83–99% of the variance in log fitness, depending on the mean selection coefficient. Ultra-rare variants (<0.01%) contribute a substantial fraction of this.
Biological Plausibility: The model predicts that the best individual in the population has an expected number of viable offspring ( $W_b$ ) of less than 2.2 to 3.0, which is consistent with human life history constraints (upper bound < 6.5), resolving the mutation load paradox.
Cross-Species Applicability: High variation in load is predicted for other mammals with large mean selection coefficients and high mutation rates (e.g., mice), but not for species like Drosophila where mean selection coefficients are much smaller.
Drift Impact: Contrary to some previous claims, the inclusion of drift (and the probability of monomorphism) modestly decreases the variation in load compared to deterministic mutation-selection balance models.

Significance
The paper argues that rare, unconditionally deleterious mutations are a major source of variation in human fitness and disease susceptibility. This high variance in genetic load helps explain why heritability observed in pedigree studies is often missing from genome-wide association studies (GWAS) that rely on common variants. The authors suggest that identifying individuals at high risk of disease may require methods capable of scoring rare and ultra-rare variants, such as variant-effect prediction from whole-genome sequencing, rather than relying solely on polygenic risk scores derived from common variants. The findings support the "omnigenic" model, where mutations in peripheral genes contribute diffusely to disease risk. The authors also emphasize the ethical importance of understanding genetic load to promote equitable societies, explicitly condemning historical misuses of genetic load concepts for discriminatory policies.

The Invisible Backpacks We All Carry

The Great Genetic Backpack Race

The Power of the Rare

What This Means for the Future

More like this