Genetics

This post looks at three recent MS studies that all read metabolism off single-cell data they largely did not generate, and compares how much independent evidence each stacks on top of the borrowed atlas. Park and colleagues reprogrammed patient fibroblasts into induced neural stem cells with a protocol that preserves the donor epigenome, found hypomethylation of lipid metabolism, inflammation and interferon genes across 28 million CpG sites in both fibroblasts and derived lines, and used single-cell and single-nucleus sequencing with chromatin accessibility to find patient-enriched clusters carrying cholesterol biosynthesis and oxidative phosphorylation alongside senescence; they then used published post-mortem and spatial datasets at the validation step, finding roughly 2.3-fold more senescent radial glia-like cells in chronic active lesions and under 5 per cent in controls, with the senolytic ABT-263 blocking paracrine transfer. Soler-Sáez and colleagues made reuse the whole design, screening GEO, ArrayExpress and UCSC Cell Browser for human MS datasets with sex recorded, analysing three plus one for validation, and returning gene-level sex differences including electron transport chain genes NDUFA13, NDUFA4, UQCR10, COX6B1, COX6C, COX7C, UQCRB, UQCRH and ATP5MG raised in male adaptive immune cells, female CD8 T cells favouring oxidative phosphorylation and mitochondrial maintenance, and glycolytic LDHA and TPI1 in progressive male signatures, with the authors' own hedging intact. Li and colleagues kept only the computation, combining network pharmacology, machine learning, single-cell analysis, SHAP, Mendelian randomisation and molecular docking to propose caspase-3 as a target of L-isoleucine, aromatic lactic acid derivatives, 3-hydroxyphenethyl alcohol and D-xylose, closing with a request for experimental validation. The common output across all three is a list of named genes, and none of them measured a metabolite in MS tissue.