Many complex disease-associated genetic variants cause disease indirectly by altering intermediate molecular phenotypes like gene expression. Statistical models can find associations between genetic variants and gene expression (i.e., expression Quantitative Trait Loci; eQTL). However, disease-relevant changes in expression often occur in specific cellular contexts (e.g., specific celltypes), which are easily obscured in bulk sequencing data. Single-cell sequencing technologies promise to overcome these limitations and improve our understanding of the mechanisms of diseases, especially those that manifest in complex organs like the lung.
We performed single-cell RNA-sequencing on half a million cells from lung tissue samples collected from 116 donors (67 with interstitial lung disease). We applied our optimized single-cell eQTL mapping workflow (Cuomo, Alvari, Azodi et. al 2021) to map eQTL in 38 celltypes and tested multiple approaches to improve power and effect size estimates by sharing information across celltypes (e.g., multivariate adaptive shrinkage). Nearly a third of the 5,174 cis-eQTL were found to have celltype dependent effects. Our results reveal new insights into the cellular context of known lung disease relevant eQTL. For example, we demonstrate that a previously identified eQTL for desmoplakin, a specialized adhesive protein critical for wound healing, is present in alveolar, but not bronchial, epidermal cells, further pinpointing the mechanistic role of desmoplakin in the pathogenesis of lung disease. We also uncovered numerous previously undescribed celltype dependent eQTL, a rich resource that could help further elucidate the genetic basis of interstitial lung disease.
Extending the impact of this work beyond lung disease, we have developed a Bioconductor container to store multi-state (e.g., multi-celltype) QTL data and a suite of tools to manipulate, analyze, and visualize multi-state QTL. This robust, common data infrastructure will improve interoperability across software and encourage the establishment of standard best practice workflows for multi-state QTL analyses.