Scalable single-cell models for robust cell-state-dependent eQTL mapping

Modelling cell state-dependent genetic associations with single-cell gene expression exhibits statistical and computational challenges. First, parametrization of single-cell gene expression profiles is not a straightforward task because individual genes exhibit distinct distributions. Second, current single-cell datasets consist of hundreds of thousands to millions of cells, which constrains the ability to test associations in a scalable manner. In this talk, I will introduce a new generalizable approach to robustly identify cell state-dependent eQTLs in single-cell data. To overcome the challenge of gene expression parametrization, we implemented a non-parametric bootstrap procedure to compute empirically calibrated p-values for variant-gene expression associations. To speed up the computation, we used the Julia programming language and pre-computed covariate-adjusted gene expression profiles with a linear mixed model before testing cell state-dependent eQTL interactions. Finally, I will demonstrate an application of this approach to identify autoimmune disease risk loci with context-specific effects in memory T cells.