NBSR: a Negative Binomial Softmax Regression model for microRNA-seq data analysis.

MicroRNAs play a central role in the regulation of gene expression and the modulation of diseases. Despite their importance, statistical methods for analyzing microRNAs have received less attention compared to messenger RNAs. Critically, messenger RNA sequencing methods are often applied to analyze microRNA sequencing data without considering the unique characteristics of microRNAs. This study examines the assumptions of messenger RNA-based methods and shows that they may incur high false discovery rates. We propose a negative binomial softmax regression (NBSR) model for microRNA sequencing data. Our approach has several advantages over existing methods. First, differential expression across experimental conditions is interpreted using the log relative abundance ratio (log-RAR). Second, it achieves greater statistical power and narrower confidence intervals by modeling the relationship between the biological coefficient of variation and relative abundance. Third, NBSR effectively handles highly variable and sparsely expressed microRNAs, resulting in improved sensitivity for detecting differential expression. Additionally, we show that debiasing the log-RAR enables accurate inference of fold changes in absolute abundance, particularly when only a small subset of microRNAs differ in expression between conditions. We demonstrate the efficacy of our approach using both real and simulated data.