Polygenic risk scores (PRS) quantify the genetic liability to disease and are calculated using an individual's genotype profile and disease-specific genome-wide association study (GWAS) summary statistics. Type 1 (T1D) and type 2 (T2D) diabetes both are determined in part by genetic loci. Correctly differentiating between types of diabetes is crucial for accurate diagnosis and treatment. PRS have the potential to address possible misclassification of T1D and T2D. Here we evaluated PRS models for T1D and T2D in European genetic ancestry participants from the UK Biobank (UKB) and then in the Michigan Genomics Initiative (MGI). Specifically, we investigated the utility of T1D and T2D PRS to discriminate between T1D, T2D, and controls in unrelated UKB individuals of European ancestry. We derived PRS models using external non-UKB GWAS. The T1D PRS model with the best discrimination between T1D cases and controls (area under the receiver operator curve [AUC] = 0.805) also yielded the best discrimination of T1D from T2D cases in the UKB (AUC = 0.792) and separation in MGI (AUC = 0.686). In contrast, the best T2D model did not discriminate between T1D and T2D cases (AUC = 0.527). Our analysis suggests that a T1D PRS model based on independent single nucleotide polymorphisms may help differentiate between T1D, T2D, and controls in individuals of European genetic ancestry.

中文翻译：

---

评估多基因风险评分以区分 1 型和 2 型糖尿病

多基因风险评分 (PRS) 量化了疾病的遗传倾向，并使用个体的基因型谱和疾病特异性全基因组关联研究 (GWAS) 摘要统计数据进行计算。1 型 (T1D) 和 2 型 (T2D) 糖尿病部分由基因位点决定。正确区分糖尿病类型对于准确诊断和治疗至关重要。PRS 有可能解决 T1D 和 T2D 可能的错误分类问题。在这里，我们评估了来自英国生物银行 (UKB) 和密歇根基因组计划 (MGI) 的欧洲遗传血统参与者的 T1D 和 T2D PRS 模型。具体来说，我们研究了 T1D 和 T2D PRS 在区分欧洲血统的无关 UKB 个体中的 T1D、T2D 和对照的效用。我们使用外部非 UKB GWAS 导出 PRS 模型。T1D PRS 模型在 T1D 病例和对照之间具有最佳区分度（接受者操作曲线下面积 [AUC] = 0.805），在 UKB 中也能最好地区分 T1D 与 T2D 病例（AUC = 0.792），并且在 MGI 中分离（AUC） = 0.686）。相比之下，最好的 T2D 模型不区分 T1D 和 T2D 病例 (AUC = 0.527)。我们的分析表明，基于独立单核苷酸多态性的 T1D PRS 模型可能有助于区分欧洲遗传血统个体中的 T1D、T2D 和对照。最好的 T2D 模型没有区分 T1D 和 T2D 病例 (AUC = 0.527)。我们的分析表明，基于独立单核苷酸多态性的 T1D PRS 模型可能有助于区分欧洲遗传血统个体中的 T1D、T2D 和对照。最好的 T2D 模型没有区分 T1D 和 T2D 病例 (AUC = 0.527)。我们的分析表明，基于独立单核苷酸多态性的 T1D PRS 模型可能有助于区分欧洲遗传血统个体中的 T1D、T2D 和对照。