Accessible Deep Learning for Plant Biology: Predicting Gene Expression from Regulatory DNA with deepCRE

Deciphering how genetic variation influences phenotype remains a major challenge in biology. Deep learning models trained on large genomic datasets provide a powerful framework to address key elements of that challenge, yet their application in plant systems remains under explored. Here, we use pre-trained deep learning models to predict gene expression from regulatory DNA sequences in plants. Our “deepCRE” approach accurately captures the effects of natural genetic variants on promoter activity and enables in silico predictions of designed mutations and promoter swap experiments. DeepCRE provided a scalable, data-driven strategy of investigating genotype–phenotype relationships, as demonstrated by our analysis of regulatory variant effects on RAP2.12, a key gene in abiotic stress resistance. Importantly, our method makes deep learning accessible to biologists without prior computational expertise, broadening its impact. The deepCRE toolkit is accessible at https://deepcre.ipk-gatersleben.de, with source code and a containerized version available for self-hosting.