Deep learning resolves cell‐type specific transcriptomes and unveils synaptic changes associated with neurodegeneration

In this study, we introduce a computational approach utilizing the Cellformer algorithm to deconvolute bulk RNA sequencing data into cell-type-specific profiles, enabling in-depth analysis of bulk RNAseq datasets. We demonstrate Cellformer's proficiency in recovering cell-type-specific RNA expression patterns. By comparing deconvoluted profiles between healthy and injured hippocampi, we unveil insights previously masked by the limitations of snRNA-seq, revealing intricate synaptic signaling dynamics. Cellformer offers the unprecedented ability to investigate extranuclear signaling in neurodegeneration.