Long-read sequencing (LRS) technologies have firmly established themselves as viable alternatives to short-read methods in various applications. Over the past five years, LRS platforms such as Oxford Nanopore and Pacific Biosciences have significantly improved in accuracy and throughput, becoming realistic alternatives to Illumina RNA-seq for transcriptomics experiments. Our lab’s pioneering work led to the creation of SQANTI, a widely used tool for the quality control and quantification of long-read transcript models. However, as technologies advance, the need for additional algorithms to support transcriptomics applications of LRS has grown. This includes not only accurate transcript identification and quantification but also analysis and correction of technological biases, experimental design considerations for large experiments, benchmarking resources, data visualization, and functional annotation of full-length transcript variants.

I will present the current suite of SQANTI tools, which provide a comprehensive solution to these data processing needs. For benchmarking, SQANTI-SIM is unique for simulating long-read and orthogonal data with precise control over transcript novelty. This enables the assessment of the identification of both annotated and novel transcript models. BUGSI is a set of universal single-isoform genes that can serve as an internal standard to troubleshoot RNA degradation and library preparation errors. The raw data quality in multi-sample experiments is assessed with SQANTI-reads, which identifies outliers, technology biases, and evaluates whether the data meets quality standards for discovery. SQANTI3 evaluates the performance of transcript reconstruction algorithms to accurately identify transcript models from long reads. The Filter, Rescue, and Requant modules curate transcript models to enhance transcriptome quality and quantification. Our methodologies reveal that lrRNA-seq data exhibit quantification biases distinct from those seen in short-read RNA-seq, necessitating specialized normalization methods. I will also discuss alternative approaches to defining joint transcriptomes in multi-sample experiments and their impact on transcript identification. Finally, I will introduce IsoAnnot, now integrated into the SQANTI framework, which distinguishes productive from unproductive transcripts and annotates them with functional labels to explore the biological impact of alternative splicing.