A dynamic single cell-based framework for digital twins to prioritize disease genes and drug targets

A major health care problems today is that drug treatment is ineffective for many patients. This may depend on the involvement of thousands of genes, which vary between patients with the same diagnosis at different time points. Identification and prioritization of upstream regulatory (UR) genes for precision medicine are therefore great challenges. In this study, we present a framework to model dynamic cellulome- and genome-wide changes in digital twins, to prioritize UR genes. We used seasonal allergic rhinitis (SAR) as a disease model because the external trigger (pollen) is known, so that the disease process can be mimicked by time series analyses of in vitro allergen-stimulated PBMC from SAR patients. Based on single cell profiling data from different stages of the disease, we constructed network models, hypothesizing that the directed molecular interactions between cell types could be traced to find an UR cell type and gene. This hypothesis was rejected because the cellular interactions formed multi-directional networks, rather than linear hierarchies. Instead, we were successfully able to rank and prioritize the URs based on their relative effects on the different cell types. We propose that this ranking strategy is a scalable approach for organizing and analyzing cellulome- and genome-wide data in digital twins. Data from scRNA-seq analysis of peripheral blood mononuclear cells (PBMCs), from patients with seasonal allergic rhinitis (SAR) as well as healthy controls. The samples (retrieved outside of the pollen season) were stimulated with pollen extract or diluent, for different time points, 0h (before stimulation), 12 hours, 1 day, 2 days, 3 days, 5 days, and 7 days.