Sex differences have been shown in laboratory biomarkers; however, the extent to which this is due to genetics is unknown. In this study, we infer sex-specific genetic parameters (heritability and genetic correlation) across 33 quantitative biomarker traits in 181,064 females and 156,135 males from the UK Biobank study. We apply a Bayesian Mixture Model, Sex Effects Mixture Model (SEMM), to Genome-wide Association Study summary statistics in order to (1) estimate the contributions of sex to the genetic variance of these biomarkers and (2) identify variants whose statistical association with these traits is sex-specific. We find that the genetics of most biomarker traits are shared between males and females, with the notable exception of testosterone, where we identify 119 female and 445 male-specific variants. These include protein-altering variants in steroid hormone production genes (POR, UGT2B7). Using the sex-specific variants as genetic instruments for Mendelian randomization, we find evidence for causal links between testosterone levels and height, body mass index, waist and hip circumference, and type 2 diabetes. We also show that sex-specific polygenic risk score models for testosterone outperform a combined model. Overall, these results demonstrate that while sex has a limited role in the genetics of most biomarker traits, sex plays an important role in testosterone genetics.

实验室生物标志物已显示出性别差异；然而，这在多大程度上是由于遗传原因尚不清楚。在这项研究中，我们推断了来自英国生物银行研究的 181,064 名女性和 156,135 名男性的 33 个定量生物标志物性状的性别特异性遗传参数（遗传力和遗传相关性）。我们将贝叶斯混合模型、性别效应混合模型 (SEMM) 应用于全基因组关联研究汇总统计，以便 (1) 估计性别对这些生物标志物的遗传变异的贡献和 (2) 识别其统计关联的变体具有这些特征是性别特定的。我们发现大多数生物标志物性状的遗传学在男性和女性之间是共享的，除了睾酮，我们确定了 119 种女性和 445 种男性特异性变体。POR，UGT2B7）。使用性别特异性变异作为孟德尔随机化的遗传工具，我们发现了睾酮水平与身高、体重指数、腰围和臀围以及 2 型糖尿病之间存在因果关系的证据。我们还表明，睾酮的性别特异性多基因风险评分模型优于组合模型。总体而言，这些结果表明，虽然性别在大多数生物标志物性状的遗传学中的作用有限，但性别在睾酮遗传学中起着重要作用。