Search

Expression quantitative trait loci (eQTL) studies are used to understand the regulatory function of non-coding genome-wide association study (GWAS) risk loci, but colocalization alone does not demonstrate a causal relationship of gene expression affecting a trait. Evidence for mediation, that perturbation of gene expression in a given tissue or developmental context will induce a change in the downstream GWAS trait, can be provided by two-sample Mendelian Randomization (MR). Here, we introduce a new statistical method, MRLocus, for Bayesian estimation of the gene-to-trait effect from eQTL and GWAS summary data for loci with evidence of allelic heterogeneity, that is, containing multiple causal variants. MRLocus makes use of a colocalization step applied to each nearly-LD-independent eQTL, followed by an MR analysis step across eQTLs. Additionally, our method involves estimation of the extent of allelic heterogeneity through a dispersion parameter, indicating variable mediation effects from each individual eQTL on the downstream trait. Our method is evaluated against other state-of-the-art methods for estimation of the gene-to-trait mediation effect, using an existing simulation framework. In simulation, MRLocus often has the highest accuracy among competing methods, and in each case provides more accurate estimation of uncertainty as assessed through interval coverage. MRLocus is then applied to five candidate causal genes for mediation of particular GWAS traits, where gene-to-trait effects are concordant with those previously reported. We find that MRLocus's estimation of the causal effect across eQTLs within a locus provides useful information for determining how perturbation of gene expression or individual regulatory elements will affect downstream traits. The MRLocus method is implemented as an R package available at https://mikelove.github.io/mrlocus.

Most genome-wide association and fine-mapping studies to date have been conducted in individuals of European descent, and genetic studies of populations of Hispanic/Latino and African ancestry are limited. In addition, these populations have more complex linkage disequilibrium structure. In order to better define the genetic architecture of these understudied populations, we leveraged >100,000 phased sequences available from deep-coverage whole genome sequencing through the multi-ethnic NHLBI Trans-Omics for Precision Medicine (TOPMed) program to impute genotypes into admixed African and Hispanic/Latino samples with genome-wide genotyping array data. We demonstrated that using TOPMed sequencing data as the imputation reference panel improves genotype imputation quality in these populations, which subsequently enhanced gene-mapping power for complex traits. For rare variants with minor allele frequency (MAF) < 0.5%, we observed a 2.3- to 6.1-fold increase in the number of well-imputed variants, with 11-34% improvement in average imputation quality, compared to the state-of-the-art 1000 Genomes Project Phase 3 and Haplotype Reference Consortium reference panels. Impressively, even for extremely rare variants with minor allele count 86%. Subsequent association analyses of TOPMed reference panel-imputed genotype data with hematological traits (hemoglobin (HGB), hematocrit (HCT), and white blood cell count (WBC)) in ~21,600 African-ancestry and ~21,700 Hispanic/Latino individuals identified associations with two rare variants in the HBB gene (rs33930165 with higher WBC [p = 8.8x10-15] in African populations, rs11549407 with lower HGB [p = 1.5x10-12] and HCT [p = 8.8x10-10] in Hispanics/Latinos). By comparison, neither variant would have been genome-wide significant if either 1000 Genomes Project Phase 3 or Haplotype Reference Consortium reference panels had been used for imputation. Our findings highlight the utility of the TOPMed imputation reference panel for identification of novel rare variant associations not previously detected in similarly sized genome-wide studies of under-represented African and Hispanic/Latino populations.

Abstract Background Machine learning tools such as random forests provide important opportunities for modeling large, complex modern data generated in medicine. Unfortunately, when it comes to understanding why machine learning models are predictive, applied research continues to rely on ‘out of bag’ (OOB) variable importance metrics (VIMPs) that are known to have considerable shortcomings within the statistics community. After explaining the limitations of OOB VIMPs – including bias towards correlated features and limited interpretability – we describe a modern approach called ‘knockoff VIMPs’ and explain its advantages. Methods We first evaluate current VIMP practices through an in-depth literature review of 50 recent random forest manuscripts. Next, we recommend organized and interpretable strategies for analysis with knockoff VIMPs, including computing them for groups of features and considering multiple model performance metrics. To demonstrate methods, we develop a random forest to predict 5-year incident stroke in the Sleep Heart Health Study and compare results based on OOB and knockoff VIMPs. Results Nearly all papers in the literature review contained substantial limitations in their use of VIMPs. In our demonstration, using OOB VIMPs for individual variables suggested two highly correlated lung function variables (forced expiratory volume, forced vital capacity) as the best predictors of incident stroke, followed by age and height. Using an organized analytic approach that considered knockoff VIMPs of both groups of features and individual features, the largest contributions to model sensitivity were medications (especially cardiovascular) and measured medical risk factors, while the largest contributions to model specificity were age, diastolic blood pressure, self-reported medical risk factors, polysomnography features, and pack-years of smoking. Thus, we reach very different conclusions about stroke risk factors using OOB VIMPs versus knockoff VIMPs. Conclusions The near-ubiquitous reliance on OOB VIMPs may provide misleading results for researchers who use such methods to guide their research. Given the rapid pace of scientific inquiry using machine learning, it is essential to bring modern knockoff VIMPs that are interpretable and unbiased into widespread applied practice to steer researchers using random forest machine learning toward more meaningful results.

Research studies involving human subjects require collection of and reporting on demographic data related to race and ethnicity. However, existing practices lack standardized guidelines, leading to misrepresentation and biased inferences and conclusions for underrepresented populations in research studies. For instance, sometimes there is a misconception that self-reported racial or ethnic identity may be treated as a biological variable with underlying genetic implications, overlooking its role as a social construct reflecting lived experiences of specific populations. In this manuscript, we use the We All Count data equity framework, which organizes data projects across seven stages: Funding, Motivation, Project Design, Data Collection, Analysis, Reporting, and Communication. Focusing on data collection and analysis, we use examples – both real and hypothetical – to review common practice and provide critiques and alternative recommendations. Through these examples and recommendations, we hope to provide the reader with some ideas and a starting point as they consider embedding a lens of justice, equity, diversity, and inclusivity from research conception to dissemination of findings.

Black childbearing individuals in the US experience a higher risk of postpartum weight retention (PPWR) compared to their White counterparts. Given that PPWR is related to adverse health outcomes, it is important to investigate predictors of weight-related health behaviors, such as self-weighing (i.e., using a scale at home). Regular self-weighing is an evidence-based weight management strategy, but there is minimal insight into sociodemographic factors related to frequency. The Postpartum Mothers Mobile Study (PMOMS) facilitated longitudinal ambulatory weight assessments to investigate racial inequities in PPWR. Our objective for the present study was to describe self-weighing behavior during and after pregnancy in the PMOMS cohort, as well as related demographic and psychosocial factors. Applying tree modeling and multiple regression, we examined self-weighing during and after pregnancy. Participants (N = 236) were 30.2 years old on average (SD = 4.7), with the majority being college-educated (53.8%, n = 127), earning at least $30,000 annually (61.4%, n = 145), and self-identifying as non-Hispanic White (NHW; 68.2%, n = 161). Adherence to regular self-weighing (at least once weekly) was highest among participants during pregnancy, with a considerable decline after giving birth. Low-income Black participants (earning < $30,000) were significantly less likely to reach a completion rate of ≥ 80% during pregnancy (AOR = 0.10) or the postpartum period (AOR = 0.16), compared to NHW participants earning at least $30,000 annually. Increases in perceived stress were associated with decreased odds of sustained self-weighing after delivery (AOR = 0.79). Future research should consider behavioral differences across demographic intersections, such as race and socioeconomic status, and the impact on efficacy of self-weighing.