Scholarlink

<p>Many genome-wide association studies (GWAS) conducted over the past two decades have focused on testing the one-to-one association between genetic variants and complex diseases or phenotypes. Despite their considerable success, such one-to-one testing approaches are underpowered because they do not utilize information on pleiotropic effects. Currently available public GWAS summary statistics provide resources for multi-trait association testing. Quadratic statistics are widely used to detect pleiotropic effects, as they follow a chi-square or weighted chi-square distribution under the null hypothesis. Nevertheless, such methods often fail to achieve superior power. To address this limitation, this study proposes a divided-and-combined association test (DCAT) for detecting pleiotropic effects using GWAS summary statistics. DCAT constructs the covariance matrix of <i>Z</i>-scores by combining phenotype correlation and genotype partial correlation via the Kronecker product. It consists of a family of quadratic statistics with different covariance matrix weights and leverages the respective advantages of each statistic. The distribution of each statistic is then approximated using a location-shifted generalized gamma distribution for <i>p</i>-value calculation, with the resulting <i>p</i>-values integrated via the Cauchy combination test. Simulation results demonstrate that DCAT controls the type I error at a reasonable level and achieves the highest statistical power in most scenarios. In the real-data analysis of four cardiovascular diseases, DCAT identifies more significant genes compared with the other four representative methods.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Divided-and-combined association test for pleiotropic effects with GWAS summary statistics

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月30日

查看全文
  • Boheng Hu,
  • Yaoyao Chen,
  • Xueqin Zheng,
  • Hongping Guo

摘要

Many genome-wide association studies (GWAS) conducted over the past two decades have focused on testing the one-to-one association between genetic variants and complex diseases or phenotypes. Despite their considerable success, such one-to-one testing approaches are underpowered because they do not utilize information on pleiotropic effects. Currently available public GWAS summary statistics provide resources for multi-trait association testing. Quadratic statistics are widely used to detect pleiotropic effects, as they follow a chi-square or weighted chi-square distribution under the null hypothesis. Nevertheless, such methods often fail to achieve superior power. To address this limitation, this study proposes a divided-and-combined association test (DCAT) for detecting pleiotropic effects using GWAS summary statistics. DCAT constructs the covariance matrix of Z-scores by combining phenotype correlation and genotype partial correlation via the Kronecker product. It consists of a family of quadratic statistics with different covariance matrix weights and leverages the respective advantages of each statistic. The distribution of each statistic is then approximated using a location-shifted generalized gamma distribution for p-value calculation, with the resulting p-values integrated via the Cauchy combination test. Simulation results demonstrate that DCAT controls the type I error at a reasonable level and achieves the highest statistical power in most scenarios. In the real-data analysis of four cardiovascular diseases, DCAT identifies more significant genes compared with the other four representative methods.