A UNIFIED STATISTICAL FRAMEWORK FOR SINGLE CELL AND BULK RNA SEQUENCING DATA.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) offers high resolution for gene expression but faces challenges like technical noise and "dropout" events.
• Dropout events, where RNA is not detected due to amplification failure, create false zeros and complicate data interpretation.
• Bulk RNA sequencing provides averaged expression but lacks cell-type specificity crucial for understanding complex diseases.

Purpose of the Study:

• To develop a novel computational model for analyzing both single-cell and bulk RNA sequencing data.
• To address the challenge of "dropout" events in scRNA-seq data for more accurate gene expression estimation.
• To enable precise cell type-specific gene expression profiling and deconvolution of bulk samples.

Main Methods:

• Formulation of a hierarchical model named Unified RNA-Sequencing Model (URSM).
• Integration of strengths from both single-cell and bulk RNA sequencing data.
• Empirical Bayes' approach with Expectation-Maximization (EM) algorithm and Gibbs sampling for parameter estimation and inference.

Main Results:

• URSM accurately models and corrects for dropout events in single-cell RNA sequencing data.
• The model achieves superior performance in estimating cell type-specific gene expression profiles compared to existing methods.
• URSM successfully deconvolves cell type proportions in bulk samples and imputes missing data.

Conclusions:

• The Unified RNA-Sequencing Model (URSM) provides a robust framework for analyzing complex RNA sequencing data.
• URSM enhances the accuracy of gene expression analysis by effectively handling technical noise and dropouts.
• This approach has significant implications for understanding human diseases through precise cell type-specific expression patterns.