Gene constraint scores lose predictive power for late-onset Mendelian disease, preprint finds
A bioRxiv study of 2,885 Mendelian diseases shows that both cross-species conservation and human population constraint scores decline in parallel as age of disease onset increases, with practical implications for variant prioritisation pipelines.
Researchers have posted a preprint to bioRxiv reporting that gene-level evolutionary conservation and human constraint scores — two widely used metrics for prioritising disease-causing variants — decline at comparable rates as the age of onset of Mendelian disease increases. The study assembled an age-of-onset axis for 2,885 Mendelian diseases catalogued in Orphanet and then fitted two established scores — 241-way phyloP cross-species conservation and human-population constraint derived from allele frequencies — against that axis.
The findings extend a previously demonstrated onset-dependent gradient from allele-frequency-level signals to the gene-level composite scores actually used in clinical and research variant-prioritisation workflows. In practical terms, the authors note that both phyloP and constraint scores carry progressively less signal about pathogenic variants as the phenotype manifests later in life, because purifying selection acts more weakly on genes whose effects emerge after reproductive age.
The result has direct relevance to the benchmarking of genomic language models and to the design of variant prioritisation pipelines: tools validated against early-onset disease genes may perform systematically worse when applied to adult-onset conditions. The authors suggest that onset-stratified evaluation sets should be used when assessing the performance of such models.
The work remains a preprint and has not yet been peer-reviewed.
Plain-language version
For patients, families, and general readers. Educational only — not medical advice.
Scientists studying inherited diseases often use scores that measure how well a gene has been preserved across different species, or how rarely variants in that gene appear in the general population, as clues to which gene changes might cause disease. A new study — not yet reviewed by other scientists — looked at nearly 3,000 inherited conditions and found that these scores are less useful for diseases that appear later in life. This is because natural selection has less opportunity to weed out harmful gene changes that only show up after people have already had children. The finding matters mainly for researchers building tools to identify disease-causing gene variants, particularly for conditions with adult onset. This is an educational summary, not medical advice. If anything here raises questions for you, please speak with your GP or a clinical professional.
Sources
Read the original reporting — these are the public sources this summary draws from.
-
Primary sourcePreprint bioRxiv (Cold Spring Harbor Laboratory) · 2026-10-10Gene-level evolutionary conservation and human constraint decline at comparable rates across the age of onset of Mendelian disease