Gene expression prediction under novel conditions using ATAC-seq-informed regulons

The study aims to unveil cell state transitions through the integration of single-cell RNA-sequencing and single-cell ATAC-sequencing data. Using SCENIC+ to build Gene Regulatory Networks (GRNs), we explore regulon activity as a prior for a generative deep learning model, such as Variational Autoencoders. Our goal is to improve the gene expression predicitions of an unseen perturbation. Preliminary succes in a Kaggle competition underscores our computational apporach’s potential. python_tools #per posar link