A robust method for transcript quantification with RNA-seq data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• High-throughput RNA-sequencing (RNA-seq) enables detailed transcriptome analysis.
• Accurate transcript isoform abundance estimation is crucial for differential expression analysis but faces challenges like read ambiguity and biases.

Purpose of the Study:

• To develop a robust transcript quantification method for RNA-seq data.
• To improve the accuracy of isoform abundance estimation and identify dominantly expressed transcripts.

Main Methods:

• Developed a general linear model leveraging reads spanning multiple splice junctions to enhance identifiability.
• Extended the method to simultaneously learn and correct for position-specific and sequence-specific biases.
• Utilized LASSO regression to infer a set of dominantly expressed transcripts, resolving the linear system.

Main Results:

• The proposed method demonstrated superior quantification accuracy compared to existing methods on simulated datasets.
• It effectively inferred a dominant set of expressed transcripts, unlike methods assigning expression to all candidates.
• Experimental application on real data confirmed its effectiveness for transcriptome differential analysis.

Conclusions:

• The developed method offers improved transcript quantification accuracy and dominant transcript set inference.
• It effectively addresses key challenges in RNA-seq data analysis, including read ambiguity and biases.
• This approach enhances the reliability of downstream differential gene expression studies.