Bayesian identification of differentially expressed isoforms using a novel joint model of RNA-seq data.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• RNA sequencing (RNA-seq) enables transcriptomic analysis.
• Identifying differentially expressed isoforms is crucial for understanding gene regulation.
• Existing methods may not fully capture isoform variability and differential expression.

Purpose of the Study:

• To develop a Bayesian approach, BayesIso, for identifying differentially expressed isoforms from RNA-seq data.
• To model both sample variability and differential isoform states using a joint model.
• To apply BayesIso to breast cancer data for uncovering recurrence-associated pathways.

Main Methods:

• Developed BayesIso, a Bayesian approach for isoform analysis.
• Utilized a novel joint model incorporating Poisson-Lognormal and Gamma-Gamma models for variability.
• Employed Markov Chain Monte Carlo (MCMC) for joint estimation of isoform states and model parameters.
• Validated using simulations and real RNA-seq data, including breast cancer samples.

Main Results:

• BayesIso effectively detects differentially expressed isoforms, especially for genes with multiple isoforms.
• Application to breast cancer data identified key isoforms and pathways linked to recurrence.
• Signaling pathways including PI3K/AKT/mTOR, PTEN, Jak-STAT, MAPK, and Wnt were associated with breast cancer development and recurrence.
• Upregulation of metabolism and cell cycle genes (e.g., CD36, TOP2A) and downregulation of immune response genes (e.g., NFATC1) were observed in early recurrence tumors.

Conclusions:

• BayesIso provides a robust framework for isoform-level differential expression analysis.
• The study highlights specific signaling pathways and molecular signatures associated with breast cancer recurrence.
• Identified potential biomarkers for early breast cancer recurrence, offering insights for therapeutic strategies.