Whole-genome sequencing and other studies of Alzheimer’s genetics are identifying more and more candidate regions of the genome that influence disease predisposition. However, we still lack an understanding of what these regions are doing in the context of Alzheimer’s disease, which involves a complex interplay of different biological processes and cell types. Our approach uses a combination of laboratory and computational research to disentangle this complexity. First, we used machine learning models that predict how mutations in the human genome influence different cell types. These models identified a set of mutations that seem to selectively impact blood immune cell types. Second, we conducted a set of experiments in which we synthesized hundreds of fragments of the human genome that are associated with Alzheimer’s disease, and then studied them in the context of the mouse brain and immune cells. These candidate genome fragments come from experiments conducted across the entire CIRCUITs consortium. Thus far, our results suggest that mutations that impact that function of the PU.1 gene influence Alzheimer’s disease predisposition.