Evaluating Evo 2 for plant variant effect prediction

Abstract The genomic foundation model Evo 2 enables zero-shot variant effect prediction. Here, we evaluate its performance using Arabidopsis thaliana reproductive barrier genes with experimentally confirmed gain- and loss-of-function variants, and show that Evo 2 distinguishes functionally impactful variants. Together with a sign-reversal amplitude metric that recovers a variant missed by standard scoring, these results highlight the potential of Evo 2 for causal variant prioritization in plant GWAS and QTL mapping. Competing Interest Statement The authors have declared no competing interest. Funder Information Declared Ministry of Education, Culture, Sports, Science and Technology, https://ror.org/048rj2z13, 22H05174, 21H05030, 24K01692, 23K17987, 23K14207 Suntory Foundation for Life Sciences, https://ror.org/02pkrz957, Suntory Rising Stars Encouragement Program in Life Sciences Copyright The copyright holder for this preprint is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.