Benchmarking of Computational Demultiplexing Methods for Single-Nucleus RNA Sequencing Data [dataset 2]

Single-nucleus RNA sequencing enables high-resolution profiling of complex tissues, but its high cost limits large-scale studies. Sample pooling with genetic demultiplexing is a scalable solution, yet comparative benchmarks are lacking. We benchmarked four widely used tools using variants from SNP arrays or extracted from matched bulk RNA-Seq. Performance, including accuracy, runtime, robustness, and scalability, was evaluated. Real-world application to 10x RNA-Seq from human and multi-species heart tissue demonstrates the tools' utility for demultiplexing and doublet removal. Overall design: snRNA-seq was applied to: 1) six patients for the simulated dataset (sequenced by 1-sample-per-lane); 2) four patients for the real-world dataset (sequenced by 2-sample-per-lane); 3) four patients for the application dataset (sequenced by 4-sample-per-lane); 4) 2 samples for the species-mixed dataset (1 human, 1 sheep).