Genetics

This post covers Upcott and colleagues' 2026 medRxiv preprint, which builds cell-type specific transcriptional risk scores for MS by weighting susceptibility variants through their effect on gene expression rather than by GWAS effect size. It follows the SMR, HEIDI and Bayesian colocalization pipeline against blood and brain eQTL data and single-cell atlases, the 240 eGenes of which most are new to MS, how the score is constructed by polarising eQTL betas toward the risk allele, the treatment-naive South Wales cohort chosen to avoid confounding by indication, the result that an immune transcriptional score tracks severity where neither polygenic score does, the UK Biobank brain volume validation, and the authors' own limitations.