Genetics

This post covers Strijbis, Inkster and colleagues' study in Multiple Sclerosis Journal applying sparse reduced-rank regression to glutamate-gene variants and seven MRI phenotypes in 326 MS patients from Basel and Amsterdam. It explains why a multivariate method suits correlated MRI outcomes and how the authors guarded it with Lasso and univariate models, reports the GRIN2A signal all three methods reach for brain and white matter volume, shows how the univariate and multivariate models each promote a signal the other sets aside, treats the null for HLA-DRB1*15:01 as evidence that susceptibility and disease expression may have separate genetics, sets out the glutamate and NMDA receptor biology, and closes on the confirmation work the authors identify.