Transcriptome-wide analysis of alternative splicing in women with fibromyalgia highlights GNLY as an immune candidate

Abstract

Background Fibromyalgia syndrome (FMS) affects 1 to 2 percent of the general population, with higher prevalence estimates in women. Recent evidence supports peripheral immune involvement in disease pathophysiology. Alternative splicing (AS) regulates immune cell function independently of transcript abundance and carries disease-relevant signal in related autoimmune conditions. No study has performed transcriptome-wide differential AS analysis directly in RNA-sequencing data from individuals with FMS. Methods We performed a secondary analysis of GSE221921, restricted to female participants (n = 91 FMS, n = 41 controls). Peripheral immune cell composition was estimated with ABIS, a deconvolution method specifically trained on peripheral blood mononuclear cell signatures. Differential AS was analysed with LeafCutter incorporating cell composition principal components as covariates. Differential gene expression was performed with DESeq2 under the same adjustment framework. Results FMS samples showed lower ABIS-inferred conventional monocyte estimates (Cliff's {delta} = -0.38, FDR = 0.008) and higher memory B-cell estimates (Cliff's {delta} = +0.32, FDR = 0.030). Composition-adjusted LeafCutter analysis identified 14 significant intron clusters, of which 9 were retained for biological interpretation. GNLY (granulysin) was the primary candidate, with a composition-adjusted 5' junction usage shift illustrated by pooled Sashimi visualisation. Composition adjustment substantially reduced significant differentially expressed genes from 10,803 to 3,186 (baseMean [≥] 10). Conclusions This first transcriptome-wide differential AS analysis in FMS identifies GNLY, encoding the cytolytic lymphocyte effector granulysin, as a splicing-specific candidate for functional follow-up. AS captures FMS-associated signal not detectable by gene-level expression analysis, and composition adjustment substantially altered the differential expression landscape.