ADuLT: An efficient and robust time-to-event GWAS

Abstract

Proportional hazards models have been proposed to analyse time-to-event phenotypes in genome-wide association studies (GWAS). However, little is known about the ability of proportional hazards models to identify genetic associations under different generative models and when ascertainment is present. Here we propose the age-dependent liability threshold (ADuLT) model as an alternative to a Cox regression based GWAS, here represented by SPACox. We compare ADuLT, SPACox, and standard case-control GWAS in simulations under two generative models and with varying degrees of ascertainment as well as in the iPSYCH cohort. We find Cox regression GWAS to be underpowered when cases are strongly ascertained (cases are oversampled by a factor 5), regardless of the generative model used. ADuLT is robust to ascertainment in all simulated scenarios. Then, we analyse four psychiatric disorders in iPSYCH, ADHD, Autism, Depression, and Schizophrenia, with a strong case-ascertainment. Across these psychiatric disorders, ADuLT identifies 20 independent genome-wide significant associations, case-control GWAS finds 17, and SPACox finds 8, which is consistent with simulation results. As more genetic data are being linked to electronic health records, robust GWAS methods that can make use of age-of-onset information will help increase power in analyses for common health outcomes.

Distinct explanations underlie gene-environment interactions in the UK Biobank

“…Eighth, we have analyzed British-ancestry samples from the UK Biobank, but an important future direction is to extend our analyses to cohorts of diverse genetic ancestry 52,53 , which may differ in their distributions of E variables, tagging of causal E variables by measured E variables, and/or causal GxE effects (analogous to differences in main G effects 45,54 ). Eighth, we do not analyze GxAge interaction (and we note the limited age variation in UK Biobank samples; age = 55 േ 8 years), but we highlight GxAge interaction and longitudinal data as important directions for future research 51,55,56 . Despite these limitations, our work quantifies and distinguishes three different types of GxE interaction across a broad set of traits and E variables.…”

PheWAS-based clustering of Mendelian Randomisation instruments reveals distinct mechanism-specific causal effects between obesity and educational attainment

“…Second, although it is not the most ideal handling of data, our binary traits are treated as continuous ones in our analysis. In large samples, linear and logistic regression effect estimates correlate very strongly and hence, it is likely that this choice did not impact the clustering 26 . Third, although we have attempted to minimise the arbitrary choice of parameters in our analysis, the genetic correlation threshold that determines which traits are too similar to the exposure and outcome trait is arbitrarily set at 0.75 for BMI and EDU, and could be modified but the emerging clusters may change as a consequence.…”

Associations between ADHD and risk of six psychiatric disorders: a Mendelian randomization study

“…Each analysis was restricted to the individuals in the subcohort and the cases for the specific phenotype being analyzed. Ultimately, the study found that ADuLT identifies independent genome-wide significant associations in ADHD [ 14 ].…”