Statistical modeling of isoform splicing dynamics from RNA-seq time series data

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Isoform quantification in RNA sequencing (RNA-seq) is challenging for genes with low expression or multiple isoforms.
• Existing methods do not fully leverage correlated experimental designs like time series or dosage response experiments.

Purpose of the Study:

• To introduce DICEseq, a novel method for isoform quantification specifically designed for correlated RNA-seq experiments.
• To improve the accuracy and reproducibility of isoform quantification, particularly in challenging scenarios.

Main Methods:

• DICEseq explicitly models correlations between different RNA-seq experiments.
• The method was evaluated using simulated and real RNA-seq datasets.

Main Results:

• DICEseq demonstrates superior accuracy compared to state-of-the-art methods, especially at low sequencing coverage.
• Real data analysis shows DICEseq enhances quantification reproducibility and robustness, improving estimate correlation by up to 10% for lowly expressed genes.
• The tool aids in optimizing the balance between temporal sampling and sequencing depth.

Conclusions:

• DICEseq offers a significant advancement for isoform quantification in correlated RNA-seq studies.
• The method has practical implications for experimental design and provides a robust tool for data analysis.
• Improved quantification accuracy and reproducibility are key benefits, particularly for low-expression genes.