Optimal pooled single-cell RNA-sequencing elucidates gene by environment interactions in circulating immune cells

Decoding the genetic architecture of gene expression remains one of the most important problems in human genetics. Because cellular context is often reflective of environmental exposure, mapping context-specific quantitative trait loci (QTLs) and identifying the cellular contexts that modify their effects is critical for understanding the mechanisms by which gene by environment interactions confer disease risk. Here, we introduce a compressed pooling framework for performing genetically multiplexed arrayed single-cell sequencing experiments without the need for prior genotyping. To this end, we developed freemuxlet, an extension of demuxlet, that enables genotype-free demultiplexing of pooled single-cell data applicable to multiplexed single cell RNA- and CITE-seq data. In total, we performed multiplexed CITE-seq across 384 samples from 64 donors over 12 pools in a single day. These results allowed us to generate a map of cell-type specific gene regulatory networks in response to each stimulus. Finally, by integrating imputed genotypes from the individuals, we map genetic variants associated with stimulation in each specific cell type and assess their enrichment for disease associations. Freemuxlet and clue together demonstrate the feasibility of performing population-scale gene-by-environment studies using single-cell RNA-sequencing scalable to between 100 – 1000 individuals. Peripheral blood mononuclear cells (PBMCs) from 64 individuals are incubated with 5 immunomodulatory stimulants and run using a compressed format in 12 reactions of droplet-based single-cell RNA-sequencing to yield transcriptomic and surface proteomic expression.