Quantitative isoform profiling using deep coverage long-read RNA sequencing across early endothelial differentiation

Long-read RNA-sequencing enables the profiling of full-length transcripts, but its quantification accuracy data has still not been robustly established. This is especially true for PacBio lrRNA-seq data, which were previously only available at low to moderate depth. Using a high-depth PacBio Kinnex lrRNA-seq dataset, sample-matched with Illumina short-read RNA-seq, we performed rigorous benchmarking to characterize quantification accuracy between platforms on a dataset representing differentiation of induced pluripotent stem cells into primordial endothelial cells. We identified biases impacting transcript quantification, including inferential variability within Illumina data, which can bias transcript abundance estimates for genes with complex splicing, as well as length biases in Kinnex data. Overall, PacBio and Illumina quantifications were strongly concordant, supporting that PacBio Kinnex is a reliable method for transcriptome profiling and enabling downstream biological analyses.