Torino Unveils Hidden Isoforms in Disease-Linked Gene Expression
At the ASHG 2025 meeting, researchers from the University of Chicago and Columbia University presented Torino, an innovative computational workflow designed to decode transcript structures and quantify isoform expression directly from read coverage in biobank-scale RNA-seq data. This approach challenges the reliance on pre-existing transcript annotations and opens a path to discovering unannotated, disease-relevant isoforms that traditional methods miss.
Why Isoforms Matter in Health and Disease
Genes can produce multiple isoforms through alternative splicing, intron retention, and alternative polyadenylation. These variants can have distinct functions and tissue-specific expression patterns, influencing cell identity and disease mechanisms, including cancer. However, thousands of potentially important isoforms remain hidden in plain sight because conventional RNA-seq analyses depend on annotated transcripts. Torino tackles this blind spot by inferring transcript structures and expression levels directly from sequencing data, without depending on prior annotations.
How Torino Works
Torino models RNA-seq data with Poisson non-negative matrix factorization enhanced by spatial smoothness priors. This mathematical framework allows the software to infer latent transcript structures from read coverage alone. By doing so, it reveals novel isoforms shaped by alternative splicing, intron retention, and APA that would be overlooked otherwise. The result is a transcriptome-wide view that links RNA processing events to underlying genetic architecture and disease risk.
Applying Torino to the GTEx Resource
Using the Genotype-Tissue Expression (GTEx) project as a massive testing ground, Torino analyzed 2,128 samples across 19 tissues. The method accurately recovered 18,813 GENCODE isoforms spanning 15,232 protein-coding genes while also uncovering a wealth of unannotated diversity. Among the discoveries were more than 10,000 novel cassette exon events, over 53,000 novel intron retention events, and 8,013 APA events. Many of these isoform events exhibited tissue specificity, implying functional roles in physiological contexts.
Genetic Regulation of Isoforms
By combining Torino-inferred isoform abundances with genetic data, the authors found a median of 2,829 isoform QTLs per tissue, underscoring the extensive control of RNA processing by genetic variation. Colocalization analyses with 65 GWAS traits identified 815 disease-linked variants overlapping isoQTLs. This finding suggests that unannotated splicing events could be hidden drivers of disease, offering new targets for functional studies and potential therapeutic intervention.
Insights from the Alzheimer’s Disease Brain Consortium
Turning to the brain, Torino analyzed 1,193 samples from the Alzheimer’s Disease Functional Genomics Consortium. The analysis revealed a global increase in intron retention associated with Alzheimer’s diagnosis and Braak stage, which measures NFT pathology severity. Notably, aberrant splicing was observed in AD risk genes such as PTK2B and APBB3, hinting that RNA processing defects contribute to disease progression and could serve as biomarkers or drug targets.
What This Means for the Future of Genomics
Torino demonstrates that transcriptome-wide discovery of novel, functionally important isoforms is feasible at biobank scale. By bridging RNA processing complexity with genetic architecture, Torino provides a framework for exploring regulatory dynamics across tissues, species, and diseases beyond Alzheimer’s. The study’s approach could help interpret how noncoding regulatory variation shapes disease risk through splicing and isoform diversity, ultimately informing precision medicine strategies.
Looking Ahead
Researchers anticipate expanding Torino’s reach to additional tissues and organisms, refining models of RNA processing, and integrating other layers of genomic data. As more unannotated isoforms come to light, the field moves closer to a complete map of how transcript diversity underpins health and disease, making Torino a pivotal tool for the next generation of genomic discovery.
