Generalized Multi-SNP Mediation Intersection-Union Test

Abstract

To elucidate the molecular mechanisms underlying genetic variants identified from genome-wide association studies (GWAS) for a variety of phenotypic traits encompassing binary, continuous, count, and survival outcomes, we propose a novel and flexible method to test for mediation that can simultaneously accommodate multiple genetic variants and different types of outcome variables. Specifically, we employ the intersection-union test approach combined with likelihood ratio test to detect mediation effect of multiple genetic variants via some mediator (for example, the expression of a neighboring gene) on outcome. We fit high-dimensional generalized linear mixed models under the mediation framework, separately under the null and alternative hypothesis. We leverage Laplace approximation to compute the marginal likelihood of outcome and use coordinate descent algorithm to estimate corresponding parameters. Our extensive simulations demonstrate the validity of our proposed method and substantial, up to 97%, power gains over alternative methods. Applications to real data for the study of Chlamydia trachomatis infection further showcase advantages of our method. We believe our proposed method will be of value and general interest in this post-GWAS era to disentangle the potential causal mechanism from DNA to phenotype for new drug discovery and personalized medicine.

IUSMMT: Survival mediation analysis of gene expression with multiple DNA methylation exposures and its application to cancers of TCGA

“…To ensure the desirable type I error control of IUT, we exploit P max = max( P α , P β ) to evaluate the overall significance of the mediation effect. IUT is advantageous in that the resulting mediation effect test has conceptual simplicity and feasibility of maneuver, and P max per se can be employed as the P- value for assessing the significance of mediation [ 51 , 52 , 55 , 122 ]. Obviously, when rejecting H 0 if P max is less than the given significance level of α , IUT is indeed a level- α test, representing that the type I error of IUT is guaranteed at most α once the rejection decision for H 0 is made [ 47 , 49 ].…”

“…Although it is conceptually straightforward, this naïve IUT-based approach is oftentimes extremely conservative especially in large scale mediation effect tests due to its composite null nature [ 43 , 53 ], which can be equivalently expressed as a combination of three disjoint component null hypotheses. To correct the intrinsic conservativeness of IUT, we estimated the proportion for each component null hypothesis and constructed a novel null distribution for P max by fitting a three-component mixture null distribution [ 54 ], which, in contrast to the naïve uniform null distribution of P max assumed in previous literature [ 51 , 52 , 55 ], can lead to a desirable control of family-wise error rate (FWER) or false discovery rate (FDR). We thus refer to our proposed approach framework as the intersection-union survival mixture-adjusted mediation test (IUSMMT).…”

Genetic susceptibility loci for Chlamydia trachomatis endometrial infection influence expression of genes involved in T cell function, tryptophan metabolism and epithelial integrity

“…We developed a generalized multi-SNP mediation intersection-union test to rigorously assess the effects of multiple eSNPs on ascension through their alteration of eGene expression (31). The locations and FDR of significant mediator eGenes were visualized using Circos software (32).…”

“…We applied the generalized multi-SNP mediation intersection-union test (31) to the identified cis-eQTLs with FDR < 0.1, and determined that expression of 33 eGenes was increased in women with endometrial infection (FDR < 0.2, Table 3), and expression of 48 eGenes was decreased in women with endometrial infection (FDR < 0.2, Table 4). Distribution and size of these 81 eGenes, their positions within the chromosomes and -log 10(FDR) of mediation tests were demonstrated by circos plot (Figure 1).…”