ASHG 2026 Poster: Isoform-level Expression Improves Survival Signal Selection and Prediction

less than 1 minute read

Published:

I presented our work, “Isoform-level Expression Improves Survival Signal Selection and Prediction,” at the ASHG 2026 Annual Meeting in Montréal, Canada.

We developed a penalized regression framework to investigate whether isoform-level expression improves survival-associated feature selection and restricted mean survival time (RMST) prediction compared with gene-level expression.

Using simulations based on TCGA breast cancer expression data, we found that isoform-level models improved feature selection and survival prediction, particularly when isoforms from the same gene had opposing effects.

Poster

ASHG 2026 Poster

View full poster (PDF)

Authors: Yung-Han Chang, S. Taylor Head, Sean Bresnahan, Ryan Sun, and Arjun Bhattacharya.