Estimation of Gene Expression at Isoform Level from mRNA-Seq Data by Bayesian Hierarchical Modeling

Area of Science:

• Genomics
• Bioinformatics
• Statistical Modeling

Background:

• Messenger RNA sequencing (mRNA-Seq) precisely measures transcript levels and transcriptome details.
• Current mRNA-Seq analysis often lacks isoform-level expression quantification.
• Alternative splicing events and splice junctions are key transcriptome features.

Purpose of the Study:

• To infer gene expression at the isoform level using mRNA-Seq data.
• To develop a Bayesian latent variable model for isoform expression analysis.
• To evaluate the impact of normalization techniques on isoform expression inference.

Main Methods:

• A Bayesian latent variable modeling framework was developed.
• The model infers isoform expression variability and presence/absence.
• Inference was performed under various normalization techniques.

Main Results:

• The proposed framework enables inference of isoform-specific expression.
• It allows for comparisons of isoform expression across different conditions.
• The model's performance is evaluated concerning different normalization strategies.

Conclusions:

• The Bayesian framework provides a robust method for isoform-level expression inference from mRNA-Seq.
• The model accounts for expression variability and facilitates differential expression analysis.
• This approach enhances the understanding of transcriptomic complexity and alternative splicing.