Benchmarking of single nuclei RNA-seq methods on human post-mortem brain tissue

Molecular analysis of human post-mortem brain tissue holds the promise to identify disease associated mechanisms. Single nuclei RNA-sequencing (snRNA-seq) is a powerful tool for molecular-level investigations of human brain tissue with cell type resolution. In the fast-developing field of post-mortem snRNA-seq, the samples sizes of case/control studies have drastically increased over the last years. Still, to overcome genetic variability across individuals and to investigate the many relevant brain regions that have not yet been sampled, even larger cohorts are necessary. It is thus important to benchmark snRNA-seq methods against each other on relevant tissue. We compared five such methods, 10× Genomics v3.1, 10× Genomics Flex Gene Expression, Parse Biosciences Evercode v2, PIPseq v5.0 from Fluent Biosciences (now acquired by Illumina) and Smart-seq3xpress, using fresh frozen post-mortem human forebrain tissue samples. Using tissue samples from the same three donors for all methods, our investigation revealed comparable overall technical performance among the five methods but suggests that biological variability was better captured with Smart-seq3xpress. We could not model the effect of sample quality, which limits the generalizability of our results. Thus, our study suggests that the selection of snRNA-seq method should mainly be informed by the need of specific data and practical experimental considerations such as hardware requirements, ability to multiplex, tissue quantity input requirements, and transportation of samples/tissues.