Trinculo: Bayesian and frequentist multinomial logistic regression for genome-wide association studies of multi-category phenotypes

From GWASs toward Mechanistic Understanding with Case Studies in Dermatogenetics

Many human skin diseases result from the complex interplay of genetic and environmental mechanisms that are largely unknown. GWASs have yielded insight into the genetic aspect of complex disease by highlighting regions of the genome or specific genetic variants associated with disease. Leveraging this information to identify causal genes and cell types will provide insight into fundamental biology, inform diagnostics, and aid drug discovery. However, the etiological mechanisms from genetic variant to disease are still unestablished in most cases. There now exists an unprecedented wealth of data and computational methods for variant interpretation in a functional context. It can be challenging to decide where to start owing to a lack of consensus on the best way to identify causal genetic mechanisms. This article highlights 3 key aspects of genetic variant interpretation: prioritizing causal genes, cell types, and pathways. We provide a practical overview of the main methods and datasets, giving examples from recent atopic dermatitis studies to provide a blueprint for variant interpretation. A collection of resources, including brief description and links to the packages and web tools, is provided for researchers looking to start in silico follow-up genetic analysis of associated genetic variants.

Variants in the DDX6-CXCR5 autoimmune disease risk locus influence the regulatory network in immune cells and salivary gland

Fine mapping and bioinformatic analysis of the DDX6-CXCR5 genetic risk association in Sjögren's Disease (SjD) and Systemic Lupus Erythematosus (SLE) identified five common SNPs with functional evidence in immune cell types: rs4938573, rs57494551, rs4938572, rs4936443, rs7117261. Functional interrogation of nuclear protein binding affinity, enhancer/promoter regulatory activity, and chromatin-chromatin interactions in immune, salivary gland epithelial, and kidney epithelial cells revealed cell type-specific allelic effects for all five SNPs that expanded regulation beyond effects on DDX6 and CXCR5 expression. Mapping the local chromatin regulatory network revealed several additional genes of interest, including lnc-PHLDB1-1. Collectively, functional characterization implicated the risk alleles of these SNPs as modulators of promoter and/or enhancer activities that regulate cell type-specific expression of DDX6, CXCR5, and lnc-PHLDB1-1, among others. Further, these findings emphasize the importance of exploring the functional significance of SNPs in the context of complex chromatin architecture in disease-relevant cell types and tissues.

Genome-wide association study identifies Sjögren's risk loci with functional implications in immune and glandular cells

Sjögren's disease is a complex autoimmune disease with twelve established susceptibility loci. This genome-wide association study (GWAS) identifies ten novel genome-wide significant (GWS) regions in Sjögren's cases of European ancestry: CD247, NAB1, PTTG1-MIR146A, PRDM1-ATG5, TNFAIP3, XKR6, MAPT-CRHR1, RPTOR-CHMP6-BAIAP6, TYK2, SYNGR1. Polygenic risk scores yield predictability (AUROC = 0.71) and relative risk of 12.08. Interrogation of bioinformatics databases refine the associations, define local regulatory networks of GWS SNPs from the 95% credible set, and expand the implicated gene list to >40. Many GWS SNPs are eQTLs for genes within topologically associated domains in immune cells and/or eQTLs in the main target tissue, salivary glands.

Computationally efficient whole-genome regression for quantitative and binary traits

Genome-wide association analysis of cohorts with thousands of phenotypes is computationally expensive, particularly when accounting for sample relatedness or population structure. Here we present a novel machine-learning method called REGENIE for fitting a whole-genome regression model for quantitative and binary phenotypes that is substantially faster than alternatives in multi-trait analyses while maintaining statistical efficiency. The method naturally accommodates parallel analysis of multiple phenotypes and requires only local segments of the genotype matrix to be loaded in memory, in contrast to existing alternatives, which must load genome-wide matrices into memory. This results in substantial savings in compute time and memory usage. We introduce a fast, approximate Firth logistic regression test for unbalanced case-control phenotypes. The method is ideally suited to take advantage of distributed computing frameworks. We demonstrate the accuracy and computational benefits of this approach using the UK Biobank dataset with up to 407,746 individuals.

Multiomics Investigation of Hypertension and White Matter Hyperintensity as a Source of Vascular Dementia or a Comorbidity to Alzheimer's Disease

BACKGROUND

Age-related comorbidity is common and significantly increases the burden for the healthcare of the elderly. Alzheimer's disease (AD) and hypertension are the two most prevalent age-related conditions and are highly comorbid. While hypertension is a risk factor for vascular dementia (VD), hypertension with AD (ADHyp+) is often characterized as probable vascular dementia. In the absence of imaging and other diagnostic tests, differentiating the two pathological states is difficult.

OBJECTIVE

Our goals are to (1) identify differences in CSF-based vascular dementia profiles, if any, between individuals who have AD only (ADHyp-), and individuals with ADHyp+ using CSF levels of amyloid β, tau and p-tau, and (2) compare genome-wide DNA profiles of ADHyp- and ADHyp+ with an unaffected control population.

METHOD

Genotype and clinical data were used to compare healthy controls to AD Hyp- vs. AD Hyp+. We compared the CSF biomarkers followed by evaluating genome wide profiles in three groups, and mapped SNPs to genes based on position and lowest p-value. The significant genes were examined for co-expression and known disease networks.

RESULTS

We found no differences between Aβ, tau and p-tau levels between ADHyp- and ADHyp+. We found TOMM40 to be associated with ADHyp- as expected but not with ADHyp+. Interestingly, SLC9A3R2 polymorphism was associated with ADHyp+, and significant gene expression changes were observed for neighboring genes.

CONCLUSION

Through this exploratory study using a novel cohort stratification design, we highlight the genetic differences in clinically similar phenotypes, indicating the utility of genetic profiling in aiding differential diagnosis of ADHyp+ and VD.

An overview of current population genomics methods for the analysis of whole-genome resequencing data in eukaryotes

Characterizing the population history of a species and identifying loci underlying local adaptation is crucial in functional ecology, evolutionary biology, conservation and agronomy. The constant improvement of high-throughput sequencing techniques has facilitated the production of whole genome data in a wide range of species. Population genomics now provides tools to better integrate selection into a historical framework, and take into account selection when reconstructing demographic history. However, this improvement has come with a profusion of analytical tools that can confuse and discourage users. Such confusion limits the amount of information effectively retrieved from complex genomic data sets, and impairs the diffusion of the most recent analytical tools into fields such as conservation biology. It may also lead to redundancy among methods. To address these isssues, we propose an overview of more than 100 state-of-the-art methods that can deal with whole genome data. We summarize the strategies they use to infer demographic history and selection, and discuss some of their limitations. A website listing these methods is available at www.methodspopgen.com.

Combined genetic analysis of juvenile idiopathic arthritis clinical subtypes identifies novel risk loci, target genes and key regulatory mechanisms

OBJECTIVES

Juvenile idiopathic arthritis (JIA) is the most prevalent form of juvenile rheumatic disease. Our understanding of the genetic risk factors for this disease is limited due to low disease prevalence and extensive clinical heterogeneity. The objective of this research is to identify novel JIA susceptibility variants and link these variants to target genes, which is essential to facilitate the translation of genetic discoveries to clinical benefit.

METHODS

We performed a genome-wide association study (GWAS) in 3305 patients and 9196 healthy controls, and used a Bayesian model selection approach to systematically investigate specificity and sharing of associated loci across JIA clinical subtypes. Suggestive signals were followed-up for meta-analysis with a previous GWAS (2751 cases/15 886 controls). We tested for enrichment of association signals in a broad range of functional annotations, and integrated statistical fine-mapping and experimental data to identify target genes.

RESULTS

Our analysis provides evidence to support joint analysis of all JIA subtypes with the identification of five novel significant loci. Fine-mapping nominated causal single nucleotide polymorphisms with posterior inclusion probabilities ≥50% in five JIA loci. Enrichment analysis identified RELA and EBF1 as key transcription factors contributing to disease risk. Our integrative approach provided compelling evidence to prioritise target genes at six loci, highlighting mechanistic insights for the disease biology and IL6ST as a potential drug target.

CONCLUSIONS

In a large JIA GWAS, we identify five novel risk loci and describe potential function of JIA association signals that will be informative for future experimental works and therapeutic strategies.

NRG1, PIP4K2A, and HTR2C as Potential Candidate Biomarker Genes for Several Clinical Subphenotypes of Depression and Bipolar Disorder

GSK3B, BDNF, NGF, NRG1, HTR2C, and PIP4K2A play important roles in molecular mechanisms of psychiatric disorders. GSK3B occupies a central position in these molecular mechanisms and is also modulated by psychotropic drugs. BDNF regulates a number of key aspects in neurodevelopment and synaptic plasticity. NGF exerts a trophic action and is implicated in cerebral alterations associated with psychiatric disorders. NRG1 is active in neural development, synaptic plasticity, and neurotransmission. HTR2C is another important psychopharmacological target. PIP4K2A catalyzes the phosphorylation of PI5P to form PIP2, the latter being implicated in various aspects of neuronal signal transduction. In the present study, the six genes were sequenced in a cohort of 19 patients with bipolar affective disorder, 41 patients with recurrent depressive disorder, and 55 patients with depressive episode. The study revealed a number of genetic variants associated with antidepressant treatment response, time to recurrence of episodes, and depression severity. Namely, alleles of rs35641374 and rs10508649 (NRG1 and PIP4K2A) may be prognostic biomarkers of time to recurrence of depressive and manic/mixed episodes among patients with bipolar affective disorder. Alleles of NC_000008.11:g.32614509_32614510del, rs61731109, and rs10508649 (also NRG1 and PIP4K2A) seem to be predictive biomarkers of response to pharmacological antidepressant treatment on the 28th day assessed by the HDRS-17 or CGI-I scale. In particular, the allele G of rs10508649 (PIP4K2A) may increase resistance to antidepressant treatment and be at the same time protective against recurrent manic/mixed episodes. These results support previous data indicating a biological link between resistance to antidepressant treatment and mania. Bioinformatic functional annotation of associated variants revealed possible impact for transcriptional regulation of PIP4K2A. In addition, the allele A of rs2248440 (HTR2C) may be a prognostic biomarker of depression severity. This allele decreases expression of the neighboring immune system gene IL13RA2 in the putamen according to the GTEx portal. The variant rs2248440 is near rs6318 (previously associated with depression and effects of psychotropic drugs) that is an eQTL for the same gene and tissue. Finally, the study points to several protein interactions relevant in the pathogenesis of mood disorders. Functional studies using cellular or animal models are warranted to support these results.

AB_SA: Accessory genes-Based Source Attribution - tracing the source of Salmonella enterica Typhimurium environmental strains

The partitioning of pathogenic strains isolated in environmental or human cases to their sources is challenging. The pathogens usually colonize multiple animal hosts, including livestock, which contaminate the food-production chain and the environment (e.g. soil and water), posing an additional public-health burden and major challenges in the identification of the source. Genomic data opens up new opportunities for the development of statistical models aiming to indicate the likely source of pathogen contamination. Here, we propose a computationally fast and efficient multinomial logistic regression source-attribution classifier to predict the animal source of bacterial isolates based on 'source-enriched' loci extracted from the accessory-genome profiles of a pangenomic dataset. Depending on the accuracy of the model's self-attribution step, the modeller selects the number of candidate accessory genes that best fit the model for calculating the likelihood of (source) category membership. The Accessory genes-Based Source Attribution (AB_SA) method was applied to a dataset of strains of Salmonella enterica Typhimurium and its monophasic variant (S. enterica 1,4,[5],12:i:-). The model was trained on 69 strains with known animal-source categories (i.e. poultry, ruminant and pig). The AB_SA method helped to identify 8 genes as predictors among the 2802 accessory genes. The self-attribution accuracy was 80 %. The AB_SA model was then able to classify 25 of the 29 S. enterica Typhimurium and S. enterica 1,4,[5],12:i:- isolates collected from the environment (considered to be of unknown source) into a specific category (i.e. animal source), with more than 85 % of probability. The AB_SA method herein described provides a user-friendly and valuable tool for performing source-attribution studies in only a few steps. AB_SA is written in R and freely available at https://github.com/lguillier/AB_SA.

Two-stage Bayesian GWAS of 9576 individuals identifies SNP regions that are targeted by miRNAs inversely expressed in Alzheimer's and cancer

INTRODUCTION

We compared genetic variants between Alzheimer's disease (AD) and two age-related cancers-breast and prostate -to identify single-nucleotide polymorphisms (SNPs) that are associated with inverse comorbidity of AD and cancer.

METHODS

Bayesian multinomial regression was used to compare sex-stratified cases (AD and cancer) against controls in a two-stage study. A ±500 KB region around each replicated hit was imputed and analyzed after merging individuals from the two stages. The microRNAs (miRNAs) that target the genes involving these SNPs were analyzed for miRNA family enrichment.

RESULTS

We identified 137 variants with inverse odds ratios for AD and cancer located on chromosomes 19, 4, and 5. The mapped miRNAs within the network were enriched for miR-17 and miR-515 families.

DISCUSSION

The identified SNPs were rs4298154 (intergenic), within TOMM40/APOE/APOC1, MARK4, CLPTM1, and near the VDAC1/FSTL4 locus. The miRNAs identified in our network have been previously reported to have inverse expression in AD and cancer.

Early progression to active tuberculosis is a highly heritable trait driven by 3q23 in Peruvians

Of the 1.8 billion people worldwide infected with Mycobacterium tuberculosis, 5-15% will develop active tuberculosis (TB). Approximately half will progress to active TB within the first 18 months after infection, presumably because they fail to mount an effective initial immune response. Here, in a genome-wide genetic study of early TB progression, we genotype 4002 active TB cases and their household contacts in Peru. We quantify genetic heritability ([Formula: see text]) of early TB progression to be 21.2% (standard error 0.08). This suggests TB progression has a strong genetic basis, and is comparable to traits with well-established genetic bases. We identify a novel association between early TB progression and variants located in a putative enhancer region on chromosome 3q23 (rs73226617, OR = 1.18; P = 3.93 × 10-8). With in silico and in vitro analyses we identify rs73226617 or rs148722713 as the likely functional variant and ATP1B3 as a potential causal target gene with monocyte specific function.

Rare variant association testing for multicategory phenotype

Genetic association studies have provided new insights into the genetic variability of human complex traits with a focus mainly on continuous or binary traits. Methods have been proposed to take into account disease heterogeneity between subgroups of patients when studying common variants but none was specifically designed for rare variants. Because rare variants are expected to have stronger effects and to be more heterogeneously distributed among cases than common ones, subgroup analyses might be particularly attractive in this context. To address this issue, we propose an extension of burden tests by using a multinomial regression model, which enables association tests between rare variants and multicategory phenotypes. We evaluated the type I error and the power of two burden tests, CAST and WSS, by simulating data under different scenarios. In the case of genetic heterogeneity between case subgroups, we showed an advantage of multinomial regression over logistic regression, which considers all the cases against the controls. We replicated these results on real data from Moyamoya disease where the burden tests performed better when cases were stratified according to age-of-onset. We implemented the functions for association tests in the R package "Ravages" available on Github.

Fine-mapping and functional studies highlight potential causal variants for rheumatoid arthritis and type 1 diabetes

To define potentially causal variants for autoimmune disease, we fine-mapped^1,2 76 rheumatoid arthritis (11,475 cases, 15,870 controls)³ and type 1 diabetes loci (9,334 cases, 11,111 controls)⁴. After sequencing 799 1-kilobase regulatory (H3K4me3) regions within these loci in 568 individuals, we observed accurate imputation for 89% of common variants. We defined credible sets of ≤5 causal variants at 5 rheumatoid arthritis and 10 type 1 diabetes loci. We identified potentially causal missense variants at DNASE1L3, PTPN22, SH2B3, and TYK2, and noncoding variants at MEG3, CD28-CTLA4, and IL2RA. We also identified potential candidate causal variants at SIRPG and TNFAIP3. Using functional assays, we confirmed allele-specific protein binding and differential enhancer activity for three variants: the CD28-CTLA4 rs117701653 SNP, MEG3 rs34552516 indel, and TNFAIP3 rs35926684 indel.

From genome-wide associations to candidate causal variants by statistical fine-mapping

Advancing from statistical associations of complex traits with genetic markers to understanding the functional genetic variants that influence traits is often a complex process. Fine-mapping can select and prioritize genetic variants for further study, yet the multitude of analytical strategies and study designs makes it challenging to choose an optimal approach. We review the strengths and weaknesses of different fine-mapping approaches, emphasizing the main factors that affect performance. Topics include interpreting results from genome-wide association studies (GWAS), the role of linkage disequilibrium, statistical fine-mapping approaches, trans-ethnic studies, genomic annotation and data integration, and other analysis and design issues.

Statistical methods to detect pleiotropy in human complex traits

In recent years pleiotropy, the phenomenon of one genetic locus influencing several traits, has become a widely researched field in human genetics. With the increasing availability of genome-wide association study summary statistics, as well as the establishment of deeply phenotyped sample collections, it is now possible to systematically assess the genetic overlap between multiple traits and diseases. In addition to increasing power to detect associated variants, multi-trait methods can also aid our understanding of how different disorders are aetiologically linked by highlighting relevant biological pathways. A plethora of available tools to perform such analyses exists, each with their own advantages and limitations. In this review, we outline some of the currently available methods to conduct multi-trait analyses. First, we briefly introduce the concept of pleiotropy and outline the current landscape of pleiotropy research in human genetics; second, we describe analytical considerations and analysis methods; finally, we discuss future directions for the field.

Resolving TYK2 locus genotype-to-phenotype differences in autoimmunity

Thousands of genetic variants have been identified, which contribute to the development of complex diseases, but determining how to elucidate their biological consequences for translation into clinical benefit is challenging. Conflicting evidence regarding the functional impact of genetic variants in the tyrosine kinase 2 (TYK2) gene, which is differentially associated with common autoimmune diseases, currently obscures the potential of TYK2 as a therapeutic target. We aimed to resolve this conflict by performing genetic meta-analysis across disorders; subsequent molecular, cellular, in vivo, and structural functional follow-up; and epidemiological studies. Our data revealed a protective homozygous effect that defined a signaling optimum between autoimmunity and immunodeficiency and identified TYK2 as a potential drug target for certain common autoimmune disorders.

Fine-mapping inflammatory bowel disease loci to single-variant resolution

Inflammatory bowel diseases are chronic gastrointestinal inflammatory disorders that affect millions of people worldwide. Genome-wide association studies have identified 200 inflammatory bowel disease-associated loci, but few have been conclusively resolved to specific functional variants. Here we report fine-mapping of 94 inflammatory bowel disease loci using high-density genotyping in 67,852 individuals. We pinpoint 18 associations to a single causal variant with greater than 95% certainty, and an additional 27 associations to a single variant with greater than 50% certainty. These 45 variants are significantly enriched for protein-coding changes (n = 13), direct disruption of transcription-factor binding sites (n = 3), and tissue-specific epigenetic marks (n = 10), with the last category showing enrichment in specific immune cells among associations stronger in Crohn's disease and in gut mucosa among associations stronger in ulcerative colitis. The results of this study suggest that high-resolution fine-mapping in large samples can convert many discoveries from genome-wide association studies into statistically convincing causal variants, providing a powerful substrate for experimental elucidation of disease mechanisms.