Machine learning-driven identification of critical gene programs and key transcription factors in migraine.

Abstract

Background: Migraine is a complex neurological disorder characterized by recurrent episodes of severe headaches. Although genetic factors have been implicated, the precise molecular mechanisms, particularly gene expression patterns in migraine-associated brain regions, remain unclear. This study applies machine learning techniques to explore region-specific gene expression profiles and identify critical gene programs and transcription factors linked to migraine pathogenesis.

Methods: We utilized single-nucleus RNA sequencing (snRNA-seq) data from 43 brain regions, along with genome-wide association study (GWAS) data, to investigate susceptibility to migraine. The cell-type-specific expression (CELLEX) algorithm was employed to calculate specific expression profiles for each region, while non-negative matrix factorization (NMF) was applied to decompose gene programs within the single-cell data from these regions. Following the annotation of brain region expression profiles and gene programs to the genome, we employed stratified linkage disequilibrium score regression (S-LDSC) to assess the associations between brain regions, gene programs, and migraine-related SNPs. Key transcription factors regulating critical gene programs were identified using a random forest model based on regulatory networks derived from the GTEx consortium.

Results: Our analysis revealed significant enrichment of migraine-associated single nucleotide polymorphisms (SNPs) in the posterior nuclear complex-medial geniculate nuclei (PoN_MG) of the thalamus, highlighting this region's crucial role in migraine pathogenesis. Gene program 1, identified through NMF, was enriched in the calcium signaling pathway, a known contributor to migraine pathophysiology. Random forest analysis predicted ARID3A as the top transcription factor regulating gene program 1, suggesting its potential role in modulating calcium-related genes involved in migraine.

Conclusion: This study provides new insights into the molecular mechanisms underlying migraine, emphasizing the importance of the PoN_MG thalamic region, calcium signaling pathways, and key transcription factors like ARID3A. These findings offer potential avenues for developing targeted therapeutic strategies for migraine treatment.