Real-time Targeted Enrichment in Single-cell Long-read Sequencing

Abstract

The vast majority of multi-exonic genes are alternatively spliced, generating diverse and cell-type-specific isoforms exhibiting functional differences. To better capture this heterogeneity using single-cell long-read sequencing data, we previously developed an exome-probe-based approach to enrich for exonic reads of target genes. While effective, this procedure is time-consuming and expensive. Real-time targeting offers a more cost-efficient solution for selectively sequencing reads of interest. Here, we performed real-time enrichment of exonic sequences of single-cell long reads by targeting spliced transcripts from 3377 genes implicated in brain functions and related diseases. Our approach increased the total number of spliced on-target reads to up to 1.82 times the control level. Notably, targeting lowly expressed subsets yielded spliced on-target reads 1.39 to 1.89 times the control. While these gains do not rival those achieved using chemical probe-based enrichment, they are sufficient to significantly enhance the power of downstream statistical analyses, such as testing for cell-type-specific isoform abundance. Specifically, compared to naïve single-cell long-read sequencing, our approach yielded 2.42 times as many genes with significant differences in isoform usage between neurons and glia. Real-time targeting confirms cell-type-specific splicing in two early Mapt exons and newly reveals such events in > 100 genes, including Bak1 and Atp8a1. Overall, our findings highlight real-time targeting as a versatile method for enhancing resolution in detecting differential isoform usage across cell types in single-cell long-read data, offering the potential to obtain a fuller view of cellular isoform diversity.