A robust mean and variance test with application to high-dimensional phenotypes

Detecting gene-environment interactions from multiple continuous traits

MOTIVATION

Genetic variants present differential effects on humans according to various environmental exposures, the so-called "gene-environment interactions" (GxE). Many diseases can be diagnosed with multiple traits, such as obesity, diabetes, and dyslipidemia. I developed a multivariate scale test (MST) for detecting the GxE of a disease with several continuous traits. Given a significant MST result, I continued to search for which trait and which E enriched the GxE signals. Simulation studies were performed to compare MST with the univariate scale test (UST).

RESULTS

MST can gain more power than UST because of (1) integrating more traits with GxE information and (2) the less harsh penalty on multiple testing. However, if only few traits account for GxE, MST may lose power due to aggregating non-informative traits into the test statistic. As an example, MST was applied to a discovery set of 93 708 Taiwan Biobank (TWB) individuals and a replication set of 25 200 TWB individuals. From among 2 570 487 SNPs with minor allele frequencies ≥5%, MST identified 18 independent variance quantitative trait loci (P < 2.4E-9 in the discovery cohort and P < 2.8E-5 in the replication cohort) and 41 GxE signals (P < .00027) based on eight trait domains (including 29 traits).

AVAILABILITY AND IMPLEMENTATION

https://github.com/WanYuLin/Multivariate-scale-test-MST.

Gene-Environment Interactions and Gene-Gene Interactions on Two Biological Age Measures: Evidence from Taiwan Biobank Participants

PhenoAge and BioAge are two commonly used biological age (BA) measures. The author here searched for gene-environment interactions (GxE) and gene-gene interactions (GxG) on PhenoAgeAccel (age-adjusted PhenoAge) and BioAgeAccel (age-adjusted BioAge) of 111,996 Taiwan Biobank (TWB) participants, including a discovery set of 86,536 TWB2 individuals and a replication set of 25,460 TWB1 individuals. Searching for variance quantitative trait loci (vQTLs) provides a convenient way to evaluate GxE and GxG. A total of 4 nearly independent (linkage disequilibrium measure r2 < 0.01) PhenoAgeAccel-vQTLs are identified from 5,303,039 autosomal TWB2 SNPs (p < 5E-8), whereas no vQTLs are found from BioAgeAccel. These 4 PhenoAgeAccel-vQTLs (rs35276921, rs141927875, rs10903013, and rs76038336) are further replicated by TWB1 (p < 5E-8). They are located in the OR51B5, FAM234A, and AXIN1 genes. All 4 PhenoAgeAccel-vQTLs are significantly associated with PhenoAgeAccel (p < 5E-8). A phylogenetic heat map of the GxE analyses showed that smoking exacerbated the PhenoAgeAccel-vQTLs' aging effects, while higher educational attainment attenuated the PhenoAgeAccel-vQTLs' aging effects. Body mass index, chronological age, alcohol consumption, and sex do not prominently modulate PhenoAgeAccel-vQTLs' aging effects. Based on these vQTL results, rs141927875-rs35276921 interaction (p = 4.7E-61) and rs76038336-rs10903013 interaction (p = 3.3E-116) on PhenoAgeAccel are detected.

Detecting genetic effects on phenotype variability to capture gene-by-environment interactions: a systematic method comparison

Genetically associated phenotypic variability has been widely observed across organisms and traits, including in humans. Both gene-gene and gene-environment interactions can lead to an increase in genetically associated phenotypic variability. Therefore, detecting the underlying genetic variants, or variance Quantitative Trait Loci (vQTLs), can provide novel insights into complex traits. Established approaches to detect vQTLs apply different methodologies from variance-only approaches to mean-variance joint tests, but a comprehensive comparison of these methods is lacking. Here, we review available methods to detect vQTLs in humans, carry out a simulation study to assess their performance under different biological scenarios of gene-environment interactions, and apply the optimal approaches for vQTL identification to gene expression data. Overall, with a minor allele frequency (MAF) of less than 0.2, the squared residual value linear model (SVLM) and the deviation regression model (DRM) are optimal when the data follow normal and non-normal distributions, respectively. In addition, the Brown-Forsythe (BF) test is one of the optimal methods when the MAF is 0.2 or larger, irrespective of phenotype distribution. Additionally, a larger sample size and more balanced sample distribution in different exposure categories increase the power of BF, SVLM, and DRM. Our results highlight vQTL detection methods that perform optimally under realistic simulation settings and show that their relative performance depends on the phenotype distribution, allele frequency, sample size, and the type of exposure in the interaction model underlying the vQTL.

Searching for gene-gene interactions through variance quantitative trait loci of 29 continuous Taiwan Biobank phenotypes

Introduction: After the era of genome-wide association studies (GWAS), thousands of genetic variants have been identified to exhibit main effects on human phenotypes. The next critical issue would be to explore the interplay between genes, the so-called "gene-gene interactions" (GxG) or epistasis. An exhaustive search for all single-nucleotide polymorphism (SNP) pairs is not recommended because this will induce a harsh penalty of multiple testing. Limiting the search of epistasis on SNPs reported by previous GWAS may miss essential interactions between SNPs without significant marginal effects. Moreover, most methods are computationally intensive and can be challenging to implement genome-wide. Methods: I here searched for GxG through variance quantitative trait loci (vQTLs) of 29 continuous Taiwan Biobank (TWB) phenotypes. A discovery cohort of 86,536 and a replication cohort of 25,460 TWB individuals were analyzed, respectively. Results: A total of 18 nearly independent vQTLs with linkage disequilibrium measure r 2 < 0.01 were identified and replicated from nine phenotypes. 15 significant GxG were found with p-values <1.1E-5 (in the discovery cohort) and false discovery rates <2% (in the replication cohort). Among these 15 GxG, 11 were detected for blood traits including red blood cells, hemoglobin, and hematocrit; 2 for total bilirubin; 1 for fasting glucose; and 1 for total cholesterol (TCHO). All GxG were observed for gene pairs on the same chromosome, except for the APOA5 (chromosome 11)-TOMM40 (chromosome 19) interaction for TCHO. Discussion: This study provided a computationally feasible way to search for GxG genome-wide and applied this approach to 29 phenotypes.

Could driving help us to "see better"? A comparative assessment of saccadic efficiency, visual speed, and attention

BACKGROUND

This study aimed at comparing drivers' and non-drivers' results in the Adult Developmental Eye Movement with Distractors test (ADEMd) and the Useful Field of View test (UFOV).

METHODS

One hundred and twenty Spaniards (mean age 50.90 ± 17.32 years) without eye disease voluntarily participated in this cross-sectional descriptive study. Participants in a single experimental session completed a questionnaire on sociodemographic, health, eyesight, and driving information. They also performed the ADEMd and UFOV tests randomly following standardized protocols. The ADEMd is a visual-verbal test that measures saccadic efficiency and visual attention. Brown-Forsythe (B-F) tests with Games-Howell post-hoc adjustments were conducted to assess differences between groups. Groups were formed according to sex, age (young adults, adults, and older adults), and driver/non-driver for further analysis. Additionally, associations between dependent variables were assessed through Spearman's correlations.

RESULTS

Drivers obtained significantly better results in the ADEMd compared with non-drivers. Non-significant differences between drivers and non-drivers were encountered in the UFOV. Additionally, significant differences were observed between sexes and age groups. It is worth highlighting that non-driver's age significantly correlated with worse ADEMd performance (rho = .637 to .716). This correlation was non-significant in drivers. Similarly, reading hours significantly correlated with better ADEMd performance in non-drivers (rho = - .291 to - .363), but not in drivers. The only significant correlations between ADEMd and UFOV tests were found in drivers (rho = .307 to .410).

CONCLUSION

Considering all the discussed results, it could be hypothesized that the driving task promotes abilities, such as oculomotor and cognitive function, which are relevant for the performance in the ADEMd. However, this hypothesis is based on correlational outcomes and further studies should causally assess this possible relation.

Assessing Differential Variability of High-Throughput DNA Methylation Data

PURPOSE OF REVIEW

DNA methylation (DNAm) is essential to human development and plays an important role as a biomarker due to its susceptibility to environmental exposures. This article reviews the current state of statistical methods developed for differential variability analysis focusing on DNAm data.

RECENT FINDINGS

With the advent of high-throughput technologies allowing for highly reliable and cost-effective measurements of DNAm, many epigenome studies have analyzed DNAm levels to uncover biological mechanisms underlying past environmental exposures and subsequent health outcomes. These studies typically focused on detecting sites or regions which differ in their mean DNAm levels among exposure groups. However, more recent studies highlighted the importance of identifying differentially variable sites or regions as biologically relevant features. Currently, the analysis of differentially variable DNAm sites has not yet gained widespread adoption in environmental studies; yet, it is important to examine the effects of environmental exposures on inter-individual epigenetic variability. In this article, we describe six of the most widely used statistical approaches for analyzing differential variability of DNAm levels and provide a discussion of their advantages and current limitations.

The impact of low input DNA on the reliability of DNA methylation as measured by the Illumina Infinium MethylationEPIC BeadChip

DNA methylation (DNAm) is commonly assayed using the Illumina Infinium MethylationEPIC BeadChip, but there is currently little published evidence to define the lower limits of the amount of DNA that can be used whilst preserving data quality. Such evidence is valuable for analyses utilizing precious or limited DNA sources. We used a single pooled sample of DNA in quadruplicate at three dilutions to define replicability and noise, and an independent population dataset of 328 individuals (from a community-based study including US-born non-Hispanic Black and white persons) to assess the impact of total DNA input on the quality of data generated using the Illumina Infinium MethylationEPIC BeadChip. We found that data are less reliable and more noisy as DNA input decreases to 40ng, with clear reductions in data quality; and that low DNA input is associated with a reduction in power to detect EWAS associations, requiring larger sample sizes. We conclude that DNA input as low as 40ng can be used with the Illumina Infinium MethylationEPIC BeadChip, provided quality checks and sensitivity analyses are undertaken.