Identification of differentially expressed genes in RNA-seq data via semi-rigid orthogonal sparse KL-NMTF

Background: Advances in transcriptome profiling have greatly improved the ability to study gene expression at a genome-wide level. Among the most widely used approaches, RNA sequencing (RNA-seq) offers high resolution, a wide dynamic range of gene expression levels, and reduced technical variability. A key application of transcriptomic analysis is the identification of differentially expressed genes (DEGs) under biological conditions, which is crucial for understanding molecular mechanisms, identifying biomarkers, and discovering therapeutic targets. Existing methods for detecting DEGs focus primarily on pairwise comparisons, potentially overlooking complex expression patterns across multiple conditions. Results: In this study, we propose a novel framework based on a new nonnegative matrix tri-factorization (NMTF) model for DEGs analysis. Unlike traditional approaches, NMTF allows the simultaneous analysis of multiple groups, capturing complex gene expression patterns while maintaining biological interpretability through its non-negativity constraints. By integrating additional biological information and operational constraints, our proposed framework improves the detection of biologically relevant gene expression changes, providing a more comprehensive and scalable solution for complex transcriptomic analysis. Conclusions: By enabling both multi-group and traditional pairwise analyses, our NMTF-based framework provides a robust and flexible approach for uncovering complex gene regulation mechanisms, offering new insights into transcriptomic data across diverse experimental conditions.