SUMMARY In Project 3, we propose to develop a next-generation of hybrid experimental framework that overcomes the limitations of single-cohort studies by leveraging heterogeneity between datasets andaccelerates in vitro hypothesis testing through human organoids (in collaboration with the Technical Project). In building this hybrid framework, we will develop novel computational methods to predict which genes in which cell types should be knocked out (or knocked in) and what downstream genes and pathways will change as a result. We will demonstrate the successful development of this hybrid framework by identifying the mechanisms underlying the transcriptome signatures we have identified for predicting vaccine response to influenza and predicting the risk of severe outcome in patients with viral infection. In collaboration with Project 1 and 2, we will identify how different antibodies relate to protective and detrimental host responses to viral infections and vaccinations. To achieve these goals, we will create the largest bulk and single-cell transcriptome database of viral infections and vaccinations to date, which we estimate will include >20,000 bulk transcriptome profiles from ~100 cohorts and >10,000,000 single-cell RNA-seq profiles from ~2,000 samples. We will perform systems immunology analysis using the advanced statistical and machine learning methods and computational frameworks developed in the Khatri lab applied to these large amounts of bulk and single-cell transcriptome data. These computational frameworks will leverage biological, clinical, and technical heterogeneity in these data to identify and refine immune signatures (genes, proteins, cell types). Because this process typically identifies hundreds or thousands of genes, we will apply several methods we have developed to reduce this list of genes, including greedy forward search. We will also use statistical deconvolution and disease trajectory inference to reduce the number of genes, while still be able to identify underlying pathways, cell types, and mechanisms. Finally, we will employ systematic ablation-based methods to infer directional interactions between genes in immune cell types in which they occur. Using these inferred directed associations, we will pose hypotheses that will be investigated using organoids in collaboration with Dr. Satpathy and the Technical Project. We will derive a pan-virus conserved host response gene signature and perform a similar analysis for vaccination, obesity, and diabetes datasets to derive respective gene signature. We will confirm the immune cell types that preferentially express these genes and understand whether a change in transcriptome or a change in cell proportion or both lead to observed signatures. We will identify overlapping immune signatures between infection, vaccination, obesity, and diabetes, which will further identify detrimental and protective host responses associated with increased or decrea...