SpotClean adjusts for spot swapping in spatial transcriptomics data

• Activity Number: 70 - Statistical Advances for Multi-Omics Data of Complex Diseases
• Type: Invited
• Date/Time: Monday, August 9, 2021 : 10:00 AM to 11:50 AM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Christina Kendziorski* and Zijian Ni and Jared Brown and Chitrasen Mohanty
• Companies: UW-Madison and UW-Madison and UW-Madison and UW-Madison
• Keywords: spatial RNA-seq; RNA-seq; single-cell RNA-seq; 10x visium; 10 Genomics

Abstract

Recent spatial transcriptomics experiments utilize slides containing thousands of spots with spot-specific barcodes that bind mRNA. In a 10x Visium experiment, for example, fresh-frozen (or FFPE) tissue is sectioned and placed onto a slide containing 4992 spots, with each spot containing millions of capture oligonucleotides with spatial barcodes unique to that spot. Ideally, a gene-specific unique molecular identifier (UMI) at a given spot would measure expression of that gene at that spot, and spots without tissue would show no UMIs. As we will demonstrate, this is not the case in practice. Messenger RNA bleed from nearby spots causes substantial contamination of UMI counts, an artifact we refer to as spot swapping. We propose SpotClean to adjust for spot swapping and, in doing so, to increase the sensitivity and precision with which downstream analyses are conducted.