Detection and removal of barcode swapping in single-cell RNA-seq data

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Barcode swapping is a known issue in Illumina sequencing, leading to mislabeled reads between multiplexed samples.
• The extent and impact of barcode swapping on genomic assays, particularly single-cell RNA sequencing (scRNA-seq), are not fully understood.
• Existing sequencing technologies may introduce artefactual cell libraries in droplet-based scRNA-seq.

Purpose of the Study:

• To quantify the fraction of swapped reads in plate-based scRNA-seq datasets using statistical methods.
• To investigate the relationship between barcode swapping and free barcode concentration.
• To develop a method for correcting barcode swapping artifacts in 10x Genomics experiments.

Main Methods:

• Utilized two statistical approaches to quantify mislabeled reads in two plate-based scRNA-seq datasets.
• Analyzed HiSeq 4000 sequencing data to determine the percentage of swapped reads.
• Developed and validated a novel algorithm to identify and exclude swapped molecules in 10x Genomics data.

Main Results:

• Approximately 2.5% of reads were found to be mislabeled between samples on the HiSeq 4000, which is lower than previously reported.
• No significant correlation was observed between the fraction of swapped reads and free barcode concentration.
• Demonstrated that barcode swapping can create artefactual cell libraries in droplet-based scRNA-seq.

Conclusions:

• Barcode swapping occurs at a lower rate than previously suggested on the HiSeq 4000 but can still impact genomic assays.
• The developed algorithm effectively removes barcode swapping artifacts in 10x Genomics experiments.
• This work enables the continued use of advanced sequencing platforms for scRNA-seq while mitigating barcode swapping issues.