Estimating the mutation prevalence in the ageing human proteome

Ageing is associated with a progressive accumulation of damaging mutations. Accordingly, a better understanding of the mutation prevalence in the ageing proteome is central to rationalizing many human diseases. Recent advances in artificial intelligence have enabled the accurate prediction of the effect of missense mutations at scale. Here, I test utility and limits of current genome-wide variant effect predictions to contribute towards a systems level understanding of ageing. The mutation risk as the likelihood to incur a damaging mutation is defined by combining human mutation rates with variant effect predictions. Modelling of the mutation risk thus revealed their intrinsic determinants, as well as details on the interplay between genotypic and phenotypic mutations during ageing at the systems level. Moreover, the mutation risk identified several cellular buffering strategies across different scales that mitigate the risk of damaging mutations, as well as a direct link to phenotype and disease annotations. Discussed are also current limitations in the prediction of variant effects that, if addressed, will likely unlock tremendous further opportunity towards the genome-wide modelling of how consequences of mutations propagate during ageing and ageing-related diseases at the systems level.