Selection of mutations by in silico and experimental variant effects (SIEVE): a new strategy to improve fitness in cool-season grasses

Quantitative genetics relies on associations between DNA changes and observed differences. These associations are useful to predict plants’ performance and select the most promising varieties. However, they are only correlations and cannot tell us what exact DNA changes are causing the observed differences.
In SIEVE, I will develop new ways of detecting associations which avoid the confusion between correlation and causation. Using machine learning, I will learn the impact of mutations on fitness based on sequence conservation across species. Then, I will validate my predictions by evaluating the impact of induced mutations in experimental populations. I will assess whether my predictions can explain observed differences for traits like metabolite production and survival.
My predictions about the effect of DNA changes will allow breeders to target the appropriate edits for improving the fitness of important crops, for example to increase grain yield in wheat or biomass in barley.