Sparse Simultaneous Signal Detection for Identifying Genetically Controlled Disease Genes

Abstract

Genome-wide association studies (GWAS) and differential expression analyses have had limited success in finding genes that cause complex diseases such as heart failure (HF), a leading cause of death in the United States. This paper proposes a new statistical approach that integrates GWAS and expression quantitative trait loci (eQTL) data to identify important HF genes. For such genes, genetic variations that perturb its expression are also likely to influence disease risk. The proposed method thus tests for the presence of simultaneous signals: SNPs that are associated with the gene’s expression as well as with disease. An analytic expression for the p-value is obtained, and the method is shown to be asymptotically adaptively optimal under certain conditions. It also allows the GWAS and eQTL data to be collected from different groups of subjects, enabling investigators to integrate public resources with their own data. Simulation experiments show that it can be more powerful than standard approaches and also robust to linkage disequilibrium between variants. The method is applied to an extensive analysis of HF genomics and identifies several genes with biological evidence for being functionally relevant in the etiology of HF. It is implemented in the R package ssa.

A statistical framework for cross-tissue transcriptome-wide association analysis

“…In addition, integrating eQTL information in genetic association analysis has become an effective way to bridge SNPs, genes, and complex traits. Many methods have been developed to co-localize eQTL with loci identified in GWAS to identify candidate risk genes for complex traits [8][9][10][11][12][13] . Two recent studies addressed this issue through an innovative approach that is sometimes referred to as transcriptome-wide association analysis.…”

Optimal detection of weak positive latent dependence between two sequences of multiple tests

“…GPA had the highest power under strong dependence, when there were many simultaneous signals, but D was the most powerful method under weak dependence. The proposed method was closely matched by the max test of Zhao et al [57] under weak dependence but outperformed the max test when there were more than 10 simultaneous signals.…”

“…The type I error refers to the null hypothesis of independence between T 1j and T 2j , so it does not apply to higher criticism because it does not test independence. The proposed D and the max test of Zhao et al [57] both had substantial power to detect dependence, and thus to detect signal in T 1j , even when higher criticism did not. Table 4: Empirical type I errors and powers for single-sequence signal detection for p = 10 5 tests at nominal significance level α = 0.05 over 400 replications.…”

“…It is interesting that the proposed D, which uses the simple permutation procedure described in Section 2.2, was still able to control the type I error in this setting. The only other method able to achieve this was the max test of Zhao et al [57]. Figure 5 reports the empirical powers and power curves of only those methods with proper type I error control.…”

“…Table 4: Empirical type I errors and powers for single-sequence signal detection for p = 10 5 tests at nominal significance level α = 0.05 over 400 replications. HC = higher criticism method of Donoho and Jin [14]; GPA = method of Chung et al [10]; Max = method of Zhao et al [57]; D = proposed method.…”