bayNorm: Bayesian gene expression recovery, imputation and normalization for single-cell RNA-sequencing data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA-sequencing (scRNA-seq) data analysis requires robust normalization due to technical variability, missing values, and batch effects.
• Existing methods often struggle to provide a unified approach for normalization, imputation, and batch correction.

Purpose of the Study:

• To develop an efficient and integrated Bayesian method for normalization, imputation, and batch effect correction of scRNA-seq data.
• To validate the proposed method's ability to accurately model scRNA-seq count distributions and improve downstream analyses.

Main Methods:

• Introduction of bayNorm, a Bayesian approach utilizing a binomial model for mRNA capture and empirical Bayes for prior estimation.
• Validation of the model's assumptions against real scRNA-seq data statistics.
• Application to public scRNA-seq datasets and simulated data for performance evaluation.

Main Results:

• bayNorm achieves robust imputation of missing values, generating realistic transcript distributions comparable to single-molecule FISH measurements.
• The method demonstrates improved accuracy and sensitivity in differential expression analysis.
• bayNorm effectively reduces batch effects compared to existing normalization techniques.

Conclusions:

• bayNorm offers an efficient, integrated solution for scaling normalization, imputation, and true count recovery in scRNA-seq data.
• The R package 'bayNorm' is available on Bioconductor, with analysis code provided on GitHub.