EXPLICIT-Transporter: a predictive model for dissecting the functions of Arabidopsis transporters

The Arabidopsis genome encodes over 1,700 transporter proteins, which mediate transmembrane transport of ions and metabolites to support diverse physiological processes. However, most transporters remain uncharacterized, and a systematic framework to predict their functions is lacking. We developed EXPLICIT-Transporter, a computational model that infers transporter functions by leveraging their co-expression with pathway genes. Trained on 32,803 transcriptome samples, the model accurately predicted the expression of 27,402 non-transporter genes using input profiles of 1,733 transporter genes. We also constructed a gene co-expression network, AtGGM2025, and identified 1,068 gene co-expression modules associated with distinct biological processes. For each module, EXPLICIT-Transporter identified a subset of predictor transporters whose expression optimally predicted module-wide gene expression, proposing them as functional candidates for associated pathways. Using this framework, we uncovered gene modules and transporters involved in tissue development, nutrient homeostasis, metabolic pathways, and subcellular organelle functions, recovering both known and novel candidates. This work establishes a genome-scale model for predicting transporter functions in Arabidopsis, accelerating their functional annotation.