New Tool for Uncovering Hidden RNA Isoforms in Disease Research
Researchers at the American Society for Human Genetics (ASHG) Annual Meeting have unveiled Torino, a groundbreaking computational workflow that reads RNA sequencing data in a new way. By leveraging biobank-scale RNA-seq datasets, Torino directly decodes transcript structures and expression levels from read coverage alone, bypassing the limitations of pre-existing transcript annotations. This approach opens the door to discovering thousands of unannotated isoforms that may play critical roles in disease mechanisms and tissue identity.
How Torino Works: From Reads to Latent Transcripts
Traditional isoform quantification relies on annotated transcripts, which leaves many potentially functional isoforms undetected. Torino tackles this gap with a Poisson non-negative matrix factorization framework that incorporates spatial smoothness priors. In plain terms, the method deconvolves RNA-seq read coverage into latent transcript structures without requiring prior annotations. The result is a transcriptome-wide map of novel, functionally relevant isoforms shaped by alternative splicing, intron retention, and alternative polyadenylation (APA).
Why This Matters for Gene Regulation and Disease
Isoforms can have distinct functions, regulatory patterns, and tissue-specific expression. By revealing unannotated isoforms, Torino helps researchers connect RNA processing steps to the genetic architecture underlying disease. The team applied Torino to the GTEx resource, which spans 2,128 samples across 19 tissues, and demonstrated that the method can recover a vast majority of known isoforms while uncovering extensive unannotated diversity. Among the newly detected events were thousands of cassette exons, intron retentions, and APA sites with strong tissue specificity, hinting at their potential functional roles.
Key Findings: Genetic Control of RNA Processing
Using Torino-inferred isoform abundances, the researchers found a median of 2,829 isoform QTLs per tissue, indicating widespread genetic influence on RNA processing decisions. When these isoQTLs were compared with 65 GWAS traits, 815 disease-associated variants overlapped with isoQTL signals. This overlap suggests that many disease drivers may lie in unannotated splicing events rather than changes to annotated exons alone, offering new targets for functional studies and therapeutic exploration.
Insights from Brain Samples and Alzheimer’s Disease
The team extended Torino’s analysis to 1,193 brain samples from the Alzheimer’s Disease Functional Genomics Consortium. They observed a global increase in intron retention associated with Alzheimer’s diagnosis and Braak stage, a measure of NFT pathology severity. Notably, aberrant splicing was detected in AD risk genes PTK2B and APBB3, pointing to RNA processing defects as a potential contributor to disease progression. These findings underscore the value of transcriptome-wide, annotation-free discovery in complex diseases where regulatory RNA events may drive pathology beyond traditional gene-centric views.
Beyond GTEx: Broader Implications and Future Directions
Torino’s model enables researchers to infer latent transcript structures across diverse tissues and species, offering a path to a more complete regulatory genome map. By directly decoding isoforms from read coverage, Torino bridges RNA processing complexity with genetic variation, disease risk, and cellular identity. As biobank-scale datasets continue to grow, Torino could become a standard tool for identifying unannotated but biologically meaningful isoforms that influence disease trajectories and treatment responses.
Authors and Acknowledgments
The Torino framework is developed by researchers including Yang Lin, PhD, associate professor of medicine at the University of Chicago, and Matthew Stephens, PhD, professor of statistics and human genetics at the University of Chicago. The team collaborated with colleagues at Columbia University, leveraging GTEx and Alzheimer’s Disease data to validate Torino’s capacity to reveal hidden RNA processing events with disease relevance.
Looking Ahead
As Torino matures, researchers anticipate applying it to more tissues, species, and disease contexts. The ability to map unannotated isoforms and quantify their genetic regulation could transform our understanding of how RNA processing contributes to health and disease, guiding future studies in oncology, neurodegeneration, and precision medicine.
