Souporcell3: robust demultiplexing for high-donor single-cell RNA-seq datasets

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Accurate demultiplexing of pooled single-cell RNA sequencing (scRNA-seq) data is crucial for large-scale genomic studies.
• Existing methods face challenges with increasing donor numbers due to overlapping genotypes and increased doublet formation.
• Robust methods are needed for high-dimensional, sparse data to maintain accuracy as sample complexity grows.

Purpose of the Study:

• To develop an enhanced souporcell method for accurate demultiplexing of scRNA-seq data with a higher number of donors.
• To improve robustness and scalability of demultiplexing algorithms in complex, high-donor scenarios.
• To provide a reliable tool for analyzing large-scale scRNA-seq datasets.

Main Methods:

• Utilizes 10x merge for initialization and K-Harmonic Means for robust clustering.
• Incorporates iterative refinement with reinitialization of low-quality clusters and locking of high-quality clusters.
• Evaluates performance against vireo, overclustered vireo, and original souporcell.

Main Results:

• The enhanced souporcell successfully demultiplexes up to 64 donors, significantly improving scalability.
• Achieves consistently high Adjusted Rand Index (ARI) scores across various doublet rates, demonstrating superior accuracy.
• Completely eliminates incorrectly merged clusters, outperforming existing methods in complex scenarios.

Conclusions:

• The enhanced souporcell method offers a scalable and accurate solution for demultiplexing high-donor scRNA-seq data.
• This advancement is critical for enabling larger and more complex single-cell genomics studies.
• The tool is freely available, promoting wider adoption in the research community.