A novel method for analyzing genetic association with longitudinal phenotypes

RETROSPECTIVE VARYING COEFFICIENT ASSOCIATION ANALYSIS OF LONGITUDINAL BINARY TRAITS: APPLICATION TO THE IDENTIFICATION OF GENETIC LOCI ASSOCIATED WITH HYPERTENSION

Many genetic studies contain rich information on longitudinal phenotypes that require powerful analytical tools for optimal analysis. Genetic analysis of longitudinal data that incorporates temporal variation is important for understanding the genetic architecture and biological variation of complex diseases. Most of the existing methods assume that the contribution of genetic variants is constant over time and fail to capture the dynamic pattern of disease progression. However, the relative influence of genetic variants on complex traits fluctuates over time. In this study, we propose a retrospective varying coefficient mixed model association test, RVMMAT, to detect time-varying genetic effect on longitudinal binary traits. We model dynamic genetic effect using smoothing splines, estimate model parameters by maximizing a double penalized quasi-likelihood function, design a joint test using a Cauchy combination method, and evaluate statistical significance via a retrospective approach to achieve robustness to model misspecification. Through simulations we illustrated that the retrospective varying-coefficient test was robust to model misspecification under different ascertainment schemes and gained power over the association methods assuming constant genetic effect. We applied RVMMAT to a genome-wide association analysis of longitudinal measure of hypertension in the Multi-Ethnic Study of Atherosclerosis. Pathway analysis identified two important pathways related to G-protein signaling and DNA damage. Our results demonstrated that RVMMAT could detect biologically relevant loci and pathways in a genome scan and provided insight into the genetic architecture of hypertension.

SCEBE: an efficient and scalable algorithm for genome-wide association studies on longitudinal outcomes with mixed-effects modeling

Genome-wide association studies (GWAS) using longitudinal phenotypes collected over time is appealing due to the improvement of power. However, computation burden has been a challenge because of the complex algorithms for modeling the longitudinal data. Approximation methods based on empirical Bayesian estimates (EBEs) from mixed-effects modeling have been developed to expedite the analysis. However, our analysis demonstrated that bias in both association test and estimation for the existing EBE-based methods remains an issue. We propose an incredibly fast and unbiased method (simultaneous correction for EBE, SCEBE) that can correct the bias in the naive EBE approach and provide unbiased P-values and estimates of effect size. Through application to Alzheimer's Disease Neuroimaging Initiative data with 6 414 695 single nucleotide polymorphisms, we demonstrated that SCEBE can efficiently perform large-scale GWAS with longitudinal outcomes, providing nearly 10 000 times improvement of computational efficiency and shortening the computation time from months to minutes. The SCEBE package and the example datasets are available at https://github.com/Myuan2019/SCEBE.

High-throughput and efficient multilocus genome-wide association study on longitudinal outcomes

MOTIVATION

With the emerging of high-dimensional genomic data, genetic analysis such as genome-wide association studies (GWAS) have played an important role in identifying disease-related genetic variants and novel treatments. Complex longitudinal phenotypes are commonly collected in medical studies. However, since limited analytical approaches are available for longitudinal traits, these data are often underutilized. In this article, we develop a high-throughput machine learning approach for multilocus GWAS using longitudinal traits by coupling Empirical Bayesian Estimates from mixed-effects modeling with a novel ℓ0-norm algorithm.

RESULTS

Extensive simulations demonstrated that the proposed approach not only provided accurate selection of single nucleotide polymorphisms (SNPs) with comparable or higher power but also robust control of false positives. More importantly, this novel approach is highly scalable and could be approximately >1000 times faster than recently published approaches, making genome-wide multilocus analysis of longitudinal traits possible. In addition, our proposed approach can simultaneously analyze millions of SNPs if the computer memory allows, thereby potentially allowing a true multilocus analysis for high-dimensional genomic data. With application to the data from Alzheimer's Disease Neuroimaging Initiative, we confirmed that our approach can identify well-known SNPs associated with AD and were much faster than recently published approaches (≥6000 times).

AVAILABILITY AND IMPLEMENTATION

The source code and the testing datasets are available at https://github.com/Myuan2019/EBE_APML0.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Retrospective Association Analysis of Longitudinal Binary Traits Identifies Important Loci and Pathways in Cocaine Use

Longitudinal phenotypes have been increasingly available in genome-wide association studies (GWAS) and electronic health record-based studies for identification of genetic variants that influence complex traits over time. For longitudinal binary data, there remain significant challenges in gene mapping, including misspecification of the model for phenotype distribution due to ascertainment. Here, we propose L-BRAT (Longitudinal Binary-trait Retrospective Association Test), a retrospective, generalized estimating equation-based method for genetic association analysis of longitudinal binary outcomes. We also develop RGMMAT, a retrospective, generalized linear mixed model-based association test. Both tests are retrospective score approaches in which genotypes are treated as random conditional on phenotype and covariates. They allow both static and time-varying covariates to be included in the analysis. Through simulations, we illustrated that retrospective association tests are robust to ascertainment and other types of phenotype model misspecification, and gain power over previous association methods. We applied L-BRAT and RGMMAT to a genome-wide association analysis of repeated measures of cocaine use in a longitudinal cohort. Pathway analysis implicated association with opioid signaling and axonal guidance signaling pathways. Lastly, we replicated important pathways in an independent cocaine dependence case-control GWAS. Our results illustrate that L-BRAT is able to detect important loci and pathways in a genome scan and to provide insights into genetic architecture of cocaine use.

A reliable method to determine which candidate chemotherapeutic drugs effectively inhibit tumor growth in patient-derived xenografts (PDX) in single mouse trials

PURPOSE

We report on a statistical method for grouping anti-cancer drugs (GRAD) in single mouse trials (SMT). The method assigns candidate drugs into groups that inhibit or do not inhibit tumor growth in patient-derived xenografts (PDX). It determines the statistical significance of the group assignments without replicate trials of each drug.

METHODS

The GRAD method applies a longitudinal finite mixture model, implemented in the statistical package PROC TRAJ, to analyze a mixture of tumor growth curves for portions of the same tumor in different mice, each single mouse exposed to a different drug. Each drug is classified into an inhibitory or non-inhibitory group. There are several advantages to the GRAD method for SMT. It determines that probability that the grouping is correct, uses the entire longitudinal tumor growth curve data for each drug treatment, can fit different shape growth curves, accounts for missing growth curve data, and accommodates growth curves of different time periods.

RESULTS

We analyzed data for 22 drugs for 18 human colorectal tumors provided by researchers in a previous publication. The GRAD method identified 18 drugs that were inhibitory against at least one tumor, and 10 tumors for which there was at least one inhibitory drug. Analysis of simulated data indicated that the GRAD method has a sensitivity of 84% and a specificity of 98%.

CONCLUSION

A statistical method, GRAD, can group anti-cancer drugs into those that are inhibitory and those that are non-inhibitory in single mouse trials and provide probabilities that the grouping is correct.

Longitudinal SNP-set association analysis of quantitative phenotypes

Many genetic epidemiological studies collect repeated measurements over time. This design not only provides a more accurate assessment of disease condition, but allows us to explore the genetic influence on disease development and progression. Thus, it is of great interest to study the longitudinal contribution of genes to disease susceptibility. Most association testing methods for longitudinal phenotypes are developed for single variant, and may have limited power to detect association, especially for variants with low minor allele frequency. We propose Longitudinal SNP-set/sequence kernel association test (LSKAT), a robust, mixed-effects method for association testing of rare and common variants with longitudinal quantitative phenotypes. LSKAT uses several random effects to account for the within-subject correlation in longitudinal data, and allows for adjustment for both static and time-varying covariates. We also present a longitudinal trait burden test (LBT), where we test association between the trait and the burden score in linear mixed models. In simulation studies, we demonstrate that LBT achieves high power when variants are almost all deleterious or all protective, while LSKAT performs well in a wide range of genetic models. By making full use of trait values from repeated measures, LSKAT is more powerful than several tests applied to a single measurement or average over all time points. Moreover, LSKAT is robust to misspecification of the covariance structure. We apply the LSKAT and LBT methods to detect association with longitudinally measured body mass index in the Framingham Heart Study, where we are able to replicate association with a circadian gene NR1D2.

Genome-wide association of trajectories of systolic blood pressure change

BACKGROUND

There is great interindividual variation in systolic blood pressure (SBP) as a result of the influences of several factors, including sex, ancestry, smoking status, medication use, and, especially, age. The majority of genetic studies have examined SBP measured cross-sectionally; however, SBP changes over time, and not necessarily in a linear fashion. Therefore, this study conducted a genome-wide association (GWA) study of SBP change trajectories using data available through the Genetic Analysis Workshop 19 (GAW19) of 959 individuals from 20 extended Mexican American families from the San Antonio Family Studies with up to 4 measures of SBP. We performed structural equation modeling (SEM) while taking into account potential genetic effects to identify how, if at all, to include covariates in estimating the SBP change trajectories using a mixture model based latent class growth modeling (LCGM) approach for use in the GWA analyses.

RESULTS

The semiparametric LCGM approach identified 5 trajectory classes that captured SBP changes across age. Each LCGM identified trajectory group was ranked based on the average number of cumulative years as hypertensive. Using a pairwise comparison of these classes the heritability estimates range from 12 to 94 % (SE = 17 to 40 %).

CONCLUSION

These identified trajectories are significantly heritable, and we identified a total of 8 promising loci that influence one's trajectory in SBP change across age. Our results demonstrate the potential utility of capitalizing on extant genetic data and longitudinal SBP assessments available through GAW19 to explore novel analytical methods with promising results.

Testing for Individual Differences in the Identification of Chemosignals for Fear and Happy: Phenotypic Super-Detectors, Detectors and Non-Detectors

Mood odor identification, explicit awareness of mood odor, may be an important emotion skill and part of a complex dual processing system. It has already been shown that mood odors have significant implicit effects, effects that occur without awareness. This study applies methods for examining human individual differences in the identification of chemosignals for fear and happy, important in itself, and a key to understanding the dual processing of emotion in the olfactory system. Axillary mood odors had been collected from 14 male donors during a mood induction task. Pads were collected after 12 and 24 minutes, creating two doses. Sixty -one participants (41 females) identified the mood odor chemosignals. On a single trial, participants identified 2 doses of fear, 2 doses of happy, and a sterile control. There were 15 trials. The first analysis (r_tt) showed that the population was phenotypically heterogeneous, not homogeneous, in identification accuracy. It also showed that a minimum of 10 trials was needed for test reliability. The second analysis, Growth Mixture Modeling, found three distinct groups of detectors: (1) 49.49% were consistently accurate super detectors, (2) 32.52% were accurate above chance level detectors, and (3) 17.98% were non-detectors. Bayesian Posterior Analyses showed reliability of groups at or above 98%. No differences related to mood odor valence (fear or happy), dose (collection at 12 or 24 minutes) or gender were found. Implications for further study of genetic differences, learning and function of identification are noted. It appears that many people can be reliable in explicitly identifying fear and happy mood odors but this skill is not homogeneous.

Longitudinal analytical approaches to genetic data

BACKGROUND

Longitudinal phenotypic data provides a rich potential resource for genetic studies which may allow for greater understanding of variants and their covariates over time. Herein, we review 3 longitudinal analytical approaches from the Genetic Analysis Workshop 19 (GAW19). These contributions investigated both genome-wide association (GWA) and whole genome sequence (WGS) data from odd numbered chromosomes on up to 4 time points for blood pressure-related phenotypes. The statistical models used included generalized estimating equations (GEEs), latent class growth modeling (LCGM), linear mixed-effect (LME), and variance components (VC). The goal of these analyses was to test statistical approaches that use repeat measurements to increase genetic signal for variant identification.

RESULTS

Two analytical methods were applied to the GAW19: GWA using real phenotypic data, and one approach to WGS using 200 simulated replicates. The first GWA approach applied a GEE-based model to identify gene-based associations with 4 derived hypertension phenotypes. This GEE model identified 1 significant locus, GRM7, which passed multiple test corrections for 2 hypertension-derived traits. The second GWA approach employed the LME to estimate genetic associations with systolic blood pressure (SBP) change trajectories identified using LCGM. This LCGM method identified 5 SBP trajectories and association analyses identified a genome-wide significant locus, near ATOX1 (p = 1.0E(-8)). Finally, a third VC-based model using WGS and simulated SBP phenotypes that constrained the β coefficient for a genetic variant across each time point was calculated and compared to an unconstrained approach. This constrained VC approach demonstrated increased power for WGS variants of moderate effect, but when larger genetic effects were present, averaging across time points was as effective.

CONCLUSION

In this paper, we summarize 3 GAW19 contributions applying novel statistical methods and testing previously proposed techniques under alternative conditions for longitudinal genetic association. We conclude that these approaches when appropriately applied have the potential to: (a) increase statistical power; (b) decrease trait heterogeneity and standard error;