vafSOMA: Detecting cell-layer-fixed somatic mutations in plants

Abstract

Somatic mutations are a source of heritable genetic variation in plants and have long been exploited by breeders through selection and vegetative propagation of bud sports. They often arise in specific meristem layers and can become fixed as descendant cell lineages expand during development. Bulk sequencing of plant organs mixes DNA from these layers, so the expected variant allele frequency of a layer-fixed mutation depends on that layer's tissue contribution, which complicates their detection. Here we present vafSOMA, a statistical model that uses tissue composition to detect fixed somatic mutations, assign them to layers and obtain counts for mutation-rate estimation. It accommodates bulk and layer-enriched samples, including those that retain DNA from other layers after enrichment. In simulations, vafSOMA performed close to the best variant allele frequency cutoff chosen in hindsight, without requiring a user-selected cutoff. Its diagnostics flagged layers for which the data did not support separate mutation counts or rates. Analysing replicate organs substantially improved detection of mutations fixed in the branch meristem and yielded estimates close to the simulated stem-cell mutation rate. Application to apricot sequencing data showed that most reference mutations fixed in the dominant layer were recovered from fruit mixtures at high depth, whereas epidermal mutations remained unresolvable in fruit and leaves. These findings suggest that published bulk somatic sequencing studies mainly capture mutations fixed in the dominant tissue layer and may therefore underestimate the combined input of fixed mutations by roughly two- to threefold when layer-specific rates are similar. vafSOMA is available at https://github.com/jlab-code/vafSOMA.