MSIQ: JOINT MODELING OF MULTIPLE RNA-SEQ SAMPLES FOR ACCURATE ISOFORM QUANTIFICATION.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• RNA sequencing (RNA-seq) is crucial for high-throughput analysis of full-length RNA isoform abundance and gene expression.
• Accurate isoform quantification is challenging due to information loss during sequencing.
• Existing methods for multiple RNA-seq samples often ignore sample quality heterogeneity, leading to biased results.

Purpose of the Study:

• To develop a novel method for accurate and robust isoform quantification by integrating multiple RNA-seq samples.
• To address the limitations of existing methods that pool samples or assign equal weights, failing to account for sample quality variations.
• To improve the understanding of gene expression regulation and transcriptome structures through enhanced quantification.

Main Methods:

• Developed a Bayesian framework called "joint modeling of multiple RNA-seq samples for accurate isoform quantification" (MSIQ).
• MSIQ identifies a consistent group of samples with homogeneous quality.
• Jointly models multiple RNA-seq samples, assigning higher weights to the consistent group for improved accuracy.

Main Results:

• MSIQ provides a consistent estimator of isoform abundance.
• Simulation studies on *D. melanogaster* genes demonstrate MSIQ's accuracy and effectiveness compared to alternative methods.
• Application studies on human embryonic stem cells, brain tissues, and HepG2 cells validate MSIQ's advantages over existing approaches.

Conclusions:

• MSIQ offers a more accurate and robust approach to isoform quantification by leveraging multiple RNA-seq samples.
• The method effectively handles sample heterogeneity, improving the reliability of gene expression analysis.
• MSIQ provides valuable insights into the impact of sample quality and experimental protocols on quantification accuracy.