Integrating short-read and long-read single-cell RNA sequencing for comprehensive transcriptome profiling in mouse retina

Abstract

The vast majority of protein-coding genes in the human genome produce multiple mRNA isoforms through alternative splicing, significantly enhancing the complexity of the transcriptome and proteome. To establish an efficient method for characterizing transcript isoforms within tissue samples, we conducted a systematic comparison between single-cell long-read and conventional short-read RNA sequencing techniques. The transcriptome of approximately 30,000 mouse retina cells was profiled using 1.54 billion Illumina short reads and 1.40 billion Oxford Nanopore Technologies long reads. Consequently, we identify 44,325 transcript isoforms, with a notable 38% previously uncharacterized and 17% expressed exclusively in distinct cellular subclasses. We observe that long-read sequencing not only matches the gene expression and cell-type annotation performance of short-read sequencing but also excel in the precise identification of transcript isoforms. While transcript isoforms are often shared across various cell types, their relative abundance shows considerable cell type–specific variation. The data generated from our study significantly enhance the existing repertoire of transcript isoforms, thereby establishing a resource for future research into the mechanisms and implications of alternative splicing within retinal biology and its links to related diseases.