Analysis of miRNA Expression Profiles Using an Integrated DESeq2 and sPLS-DA Approach to Identify Potential Multiple Sclerosis Biomarkers

Abstract

Multiple sclerosis (MS) is a complex autoimmune disorder driven by diverse molecular, genetic, and immunological processes, and characterized by diverse clinical course forms. The identification of reliable biomarkers is essential for improving our understanding of MS disease mechanisms and for developing more advanced diagnostic and prognostic strategies. microRNA expression profiling represents one of the most promising approaches for revealing molecular alterations associated with MS pathogenesis. The aim of this study was to identify potential MS biomarkers by applying sequencingbased expression analysis and machine-learning methods using untreated samples. MiRNA sequencing was performed on samples from 9 MS patients and 13 healthy controls. Following data quality control, principal component analysis (PCA) was applied to evaluate sample structure and group separation. Differential expression analysis was conducted using DESeq2, with results visualized through a volcano plot and heatmap. In parallel, sPLS-DA was employed to identify features contributing most strongly to group discrimination. Molecules overlapping between DESeq2 and sPLS-DA results were further evaluated using boxplots and univariate logistic regression models. PCA analysis showed a clear separation between the MS and control groups based on miRNA expression. DESeq2 identified miRNAs that were significantly differentially expressed. sPLS-DA further narrowed down the set of features with strong ability to distinguish the groups. All miRNAs selected by sPLS-DA overlapped with those found by DESeq2, supporting their stability and biological relevance. Logistic regression showed that even single miRNAs from this set could significantly separate the two groups. The integrated DESeq2 and sPLS-DA analysis identified a group of miRNAs that were statistically different between MS patients and healthy controls and showed strong discriminatory ability. The overlapping miRNAs represent strong candidates for potential diagnostic or prognostic MS biomarkers. The analysis was performed using untreated samples, making the results more reliable and providing a solid basis for future validation studies.