LINKING CELL STATES TO THE HERITABILITY OF DEPRESSION, COMBINING A NOVEL SINGLE-CELL ANALYSIS WITH POPULATION LEVEL DATA

Abstract

When interrogating single-cell datasets, we are interested both in what cells are, as well as what they do. Cell states represent dynamic behaviour and activities including cell cycle stage, level of maturity, response to stimuli, and spatial location. It can be difficult to disentangle states from cell type using traditional clustering approaches but STATOR, a methodology we developed recently, allows for finer resolution of cell type, subtype and state. STATOR can identify states within cell types that appear to be homogenous when clustered and displayed on two-dimensional PCA or UMAP plot. Once these states have been identified, further downstream analysis is required to understand the biological function of these cells and their contribution to disease.

The Major Depressive Disorder (MDD) working group of the PGC has recently made their MDD3 GWAS results available. This work represents the largest and most diverse GWAS study to date, providing a powerful population level tool for investigating the impact of individual variants to incidence of MDD. In addition, case vs control single-cell datasets for men and women with and without MDD, taken from the dorsolateral prefrontal cortex, have recently been published. By using published single-cell case vs control MDD datasets, we have identified cell states enriched in MDD. We then performed differential gene expression on cells with and without identified states. We then used the regions around these genes to perform stratified LD score regression using the MDD3 GWAS results.

Linking individual cell states, along with the cell types and sub-types, to population level MDD data can provide insights into what functional role sequence variation may be having at a cellular level. Determining the mechanism by which variant can be linked to trait can improve our understanding of the underlying molecular processes involved in the incidence of MDD, as well as provide opportunities for investigating potential mechanisms for intervention in the future.

While this work is presented in the context of MDD, the methodology can be used for any condition of interest where case vs control single-cell data and GWAS results are available.