Inferring metabolite states from spatial transcriptomes using multiple graph neural network

Abstract

AbstractMetabolism serves as the pivotal interface connecting genotype and phenotype. While various methods are available for predicting metabolite levels from transcriptomic data, their efficacy remains poor. We developed an efficient and adaptable algorithm known as Multiple Graph-based Flux Estimation Analysis (MGFEA). By leveraging single-cell and spatial transcriptomic data, MGFEA enables the inference of metabolite levels. Furthermore, MGFEA can further improve the accuracy of these inferences through exploiting additional metabolome data.