Fig. 1

Methods overview. a Causal diagram (DAG) representing the assumed relationship between genetic variants for CRP (\(G_{CRP}\)), measured levels of serum CRP and BMI, and major depressive disorder (MDD). The dashed line between CRP and BMI represents a potential contribution of BMI to CRP levels. b DAG for an MVMR analysis that genetically proxies both CRP and BMI (\(G_{CRP},G_{BMI}\)) enables estimation of the direct causal effect of CRP and BMI on MDD. c Estimation of the proportion of the CRP effect mediated by BMI (p\(_m\)). d Robust MVMR to account for unmeasured pleiotropy as well as measured BMI pleiotropy. If some of the genetic variants in \(G_{CRP}\) or \(G_{BMI}\) affect MDD directly, other than just through changing CRP or BMI levels, the estimated effects will be biased. Robust methods such as MR GRAPPLE protect against this