Nieuwboer Harold A, Pool René, Dolan Conor V, Boomsma Dorret I, Nivard Michel G
Department of Biological Psychology, VU University Amsterdam, 1081 BT Amsterdam, the Netherlands.
Department of Biological Psychology, VU University Amsterdam, 1081 BT Amsterdam, the Netherlands.
Am J Hum Genet. 2016 Oct 6;99(4):917-927. doi: 10.1016/j.ajhg.2016.07.020. Epub 2016 Sep 8.
Here we present a method of genome-wide inferred study (GWIS) that provides an approximation of genome-wide association study (GWAS) summary statistics for a variable that is a function of phenotypes for which GWAS summary statistics, phenotypic means, and covariances are available. A GWIS can be performed regardless of sample overlap between the GWAS of the phenotypes on which the function depends. Because a GWIS provides association estimates and their standard errors for each SNP, a GWIS can form the basis for polygenic risk scoring, LD score regression, Mendelian randomization studies, biological annotation, and other analyses. GWISs can also be used to boost power of a GWAS meta-analysis where cohorts have not measured all constituent phenotypes in the function. We demonstrate the accuracy of a BMI GWIS by performing power simulations and type I error simulations under varying circumstances, and we apply a GWIS by reconstructing a body mass index (BMI) GWAS based on a weight GWAS and a height GWAS. Furthermore, we apply a GWIS to further our understanding of the underlying genetic structure of bipolar disorder and schizophrenia and their relation to educational attainment. Our analyses suggest that the previously reported genetic correlation between schizophrenia and educational attainment is probably induced by the observed genetic correlation between schizophrenia and bipolar disorder and the previously reported genetic correlation between bipolar disorder and educational attainment.
在此，我们提出一种全基因组推断研究（GWIS）方法，该方法可为一个变量提供全基因组关联研究（GWAS）汇总统计量的近似值，该变量是某些表型的函数，而这些表型的GWAS汇总统计量、表型均值和协方差是可用的。无论该函数所依赖的表型的GWAS之间的样本是否重叠，都可以进行GWIS。由于GWIS为每个单核苷酸多态性（SNP）提供关联估计值及其标准误，因此GWIS可为多基因风险评分、连锁不平衡（LD）评分回归、孟德尔随机化研究、生物学注释及其他分析奠定基础。GWIS还可用于提高GWAS荟萃分析的效能，前提是各队列未测量该函数中的所有组成表型。我们通过在不同情况下进行效能模拟和I型错误模拟来证明体重指数（BMI）GWIS的准确性，并通过基于体重GWAS和身高GWAS重建体重指数GWAS来应用GWIS。此外，我们应用GWIS来加深对双相情感障碍和精神分裂症的潜在遗传结构及其与教育程度关系的理解。我们的分析表明，先前报道的精神分裂症与教育程度之间的遗传相关性可能是由观察到的精神分裂症与双相情感障碍之间的遗传相关性以及先前报道的双相情感障碍与教育程度之间的遗传相关性所导致的。
