Bayesian multiple logistic regression for case-control GWAS

An individualized Bayesian method for estimating genomic variants of hypertension

BACKGROUND

Genomic variants of the disease are often discovered nowadays through population-based genome-wide association studies (GWAS). Identifying genomic variations potentially underlying a phenotype, such as hypertension, in an individual is important for designing personalized treatment; however, population-level models, such as GWAS, may not capture all the important, individualized factors well. In addition, GWAS typically requires a large sample size to detect the association of low-frequency genomic variants with sufficient power. Here, we report an individualized Bayesian inference (IBI) algorithm for estimating the genomic variants that influence complex traits, such as hypertension, at the level of an individual (e.g., a patient). By modeling at the level of the individual, IBI seeks to find genomic variants observed in the individual's genome that provide a strong explanation of the phenotype observed in this individual.

RESULTS

We applied the IBI algorithm to the data from the Framingham Heart Study to explore the genomic influences of hypertension. Among the top-ranking variants identified by IBI and GWAS, there is a significant number of shared variants (intersection); the unique variants identified only by IBI tend to have relatively lower minor allele frequency than those identified by GWAS. In addition, IBI discovered more individualized and diverse variants that explain hypertension patients better than GWAS. Furthermore, IBI found several well-known low-frequency variants as well as genes related to blood pressure that GWAS missed in the same cohort. Finally, IBI identified top-ranked variants that predicted hypertension better than GWAS, according to the area under the ROC curve.

CONCLUSIONS

The results support IBI as a promising approach for complementing GWAS, especially in detecting low-frequency genomic variants as well as learning personalized genomic variants of clinical traits and disease, such as the complex trait of hypertension, to help advance precision medicine.

Discovering Genomic Regions Associated with Reproductive Traits and Frame Score in Mexican Simmental and Simbrah Cattle Using Individual SNP and Haplotype Markers

Reproductive efficiency stands as a critical determinant of profitability within beef production systems. The incorporation of molecular markers can expedite advancements in reproductive performance. While the use of SNPs in association analysis is prevalent, approaches centered on haplotypes can offer a more comprehensive insight. The study used registered Simmental and Simbrah cattle genotyped with the GGP Bovine 150 k panel. Phenotypes included scrotal circumference (SC), heifer fertility (HF), stayability (STAY), and frame score (FS). After quality control, 105,129 autosomal SNPs from 967 animals were used. Haplotype blocks were defined based on linkage disequilibrium. Comparison between haplotypes and SNPs for reproductive traits and FS was conducted using Bayesian and frequentist models. 23, 13, 7, and 2 SNPs exhibited associations with FS, SC, HF, and STAY, respectively. In addition, seven, eight, seven, and one haplotypes displayed associations with FS, SC, HF, and STAY, respectively. Within these delineated genomic segments, potential candidate genes were associated.

Local and Bayesian Survival FDR Estimations to Identify Reliable Associations in Whole Genome of Bread Wheat

Estimating the FDR significance threshold in genome-wide association studies remains a major challenge in distinguishing true positive hypotheses from false positive and negative errors. Several comparative methods for multiple testing comparison have been developed to determine the significance threshold; however, these methods may be overly conservative and lead to an increase in false negative results. The local FDR approach is suitable for testing many associations simultaneously based on the empirical Bayes perspective. In the local FDR, the maximum likelihood estimator is sensitive to bias when the GWAS model contains two or more explanatory variables as genetic parameters simultaneously. The main criticism of local FDR is that it focuses only locally on the effects of single nucleotide polymorphism (SNP) in tails of distribution, whereas the signal associations are distributed across the whole genome. The advantage of the Bayesian perspective is that knowledge of prior distribution comes from other genetic parameters included in the GWAS model, such as linkage disequilibrium (LD) analysis, minor allele frequency (MAF) and call rate of significant associations. We also proposed Bayesian survival FDR to solve the multi-collinearity and large-scale problems, respectively, in grain yield (GY) vector in bread wheat with large-scale SNP information. The objective of this study was to obtain a short list of SNPs that are reliably associated with GY under low and high levels of nitrogen (N) in the population. The five top significant SNPs were compared with different Bayesian models. Based on the time to events in the Bayesian survival analysis, the differentiation between minor and major alleles within the association panel can be identified.

Deep polygenic neural network for predicting and identifying yield-associated genes in Indonesian rice accessions

As the fourth most populous country in the world, Indonesia must increase the annual rice production rate to achieve national food security by 2050. One possible solution comes from the nanoscopic level: a genetic variant called Single Nucleotide Polymorphism (SNP), which can express significant yield-associated genes. The prior benchmark of this study utilized a statistical genetics model where no SNP position information and attention mechanism were involved. Hence, we developed a novel deep polygenic neural network, named the NucleoNet model, to address these obstacles. The NucleoNets were constructed with the combination of prominent components that include positional SNP encoding, the context vector, wide models, Elastic Net, and Shannon’s entropy loss. This polygenic modeling obtained up to 2.779 of Mean Squared Error (MSE) with 47.156% of Symmetric Mean Absolute Percentage Error (SMAPE), while revealing 15 new important SNPs. Furthermore, the NucleoNets reduced the MSE score up to 32.28% compared to the Ordinary Least Squares (OLS) model. Through the ablation study, we learned that the combination of Xavier distribution for weights initialization and Normal distribution for biases initialization sparked more various important SNPs throughout 12 chromosomes. Our findings confirmed that the NucleoNet model was successfully outperformed the OLS model and identified important SNPs to Indonesian rice yields.

MTHSA-DHEI: multitasking harmony search algorithm for detecting high-order SNP epistatic interactions

Genome-wide association studies have succeeded in identifying genetic variants associated with complex diseases, but the findings have not been well interpreted biologically. Although it is widely accepted that epistatic interactions of high-order single nucleotide polymorphisms (SNPs) [(1) Single nucleotide polymorphisms (SNP) are mainly deoxyribonucleic acid (DNA) sequence polymorphisms caused by variants at a single nucleotide at the genome level. They are the most common type of heritable variation in humans.] are important causes of complex diseases, the combinatorial explosion of millions of SNPs and multiple tests impose a large computational burden. Moreover, it is extremely challenging to correctly distinguish high-order SNP epistatic interactions from other high-order SNP combinations due to small sample sizes. In this study, a multitasking harmony search algorithm (MTHSA-DHEI) is proposed for detecting high-order epistatic interactions [(2) In classical genetics, if genes X1 and X2 are mutated and each mutation by itself produces a unique disease status (phenotype) but the mutations together cause the same disease status as the gene X1 mutation, gene X1 is epistatic and gene X2 is hypostatic, and gene X1 has an epistatic effect (main effect) on disease status. In this work, a high-order epistatic interaction occurs when two or more SNP loci have a joint influence on disease status.], with the goal of simultaneously detecting multiple types of high-order (k1-order, k2-order, …, kn-order) SNP epistatic interactions. Unified coding is adopted for multiple tasks, and four complementary association evaluation functions are employed to improve the capability of discriminating the high-order SNP epistatic interactions. We compare the proposed MTHSA-DHEI method with four excellent methods for detecting high-order SNP interactions for 8 high-orderepistatic interaction models with no marginal effect (EINMEs) and 12 epistatic interaction models with marginal effects (EIMEs) (*) and implement the MTHSA-DHEI algorithm with a real dataset: age-related macular degeneration (AMD). The experimental results indicate that MTHSA-DHEI has power and an F1-score exceeding 90% for all EIMEs and five EINMEs and reduces the computational time by more than 90%. It can efficiently perform multiple high-order detection tasks for high-order epistatic interactions and improve the discrimination ability for diverse epistasis models.

Fine-mapping studies distinguish genetic risks for childhood- and adult-onset asthma in the HLA region

BACKGROUND

Genome-wide association studies of asthma have revealed robust associations with variation across the human leukocyte antigen (HLA) complex with independent associations in the HLA class I and class II regions for both childhood-onset asthma (COA) and adult-onset asthma (AOA). However, the specific variants and genes contributing to risk are unknown.

METHODS

We used Bayesian approaches to perform genetic fine-mapping for COA and AOA (n=9432 and 21,556, respectively; n=318,167 shared controls) in White British individuals from the UK Biobank and to perform expression quantitative trait locus (eQTL) fine-mapping in immune (lymphoblastoid cell lines, n=398; peripheral blood mononuclear cells, n=132) and airway (nasal epithelial cells, n=188) cells from ethnically diverse individuals. We also examined putatively causal protein coding variation from protein crystal structures and conducted replication studies in independent multi-ethnic cohorts from the UK Biobank (COA n=1686; AOA n=3666; controls n=56,063).

RESULTS

Genetic fine-mapping revealed both shared and distinct causal variation between COA and AOA in the class I region but only distinct causal variation in the class II region. Both gene expression levels and amino acid variation contributed to risk. Our results from eQTL fine-mapping and amino acid visualization suggested that the HLA-DQA1*03:01 allele and variation associated with expression of the nonclassical HLA-DQA2 and HLA-DQB2 genes accounted entirely for the most significant association with AOA in GWAS. Our studies also suggested a potentially prominent role for HLA-C protein coding variation in the class I region in COA. We replicated putatively causal variant associations in a multi-ethnic cohort.

CONCLUSIONS

We highlight roles for both gene expression and protein coding variation in asthma risk and identified putatively causal variation and genes in the HLA region. A convergence of genomic, transcriptional, and protein coding evidence implicates the HLA-DQA2 and HLA-DQB2 genes and HLA-DQA1*03:01 allele in AOA.

From Genotype to Phenotype: Polygenic Prediction of Complex Human Traits

Decoding the genome confers the capability to predict characteristics of the organism (phenotype) from DNA (genotype). We describe the present status and future prospects of genomic prediction of complex traits in humans. Some highly heritable complex phenotypes such as height and other quantitative traits can already be predicted with reasonable accuracy from DNA alone. For many diseases, including important common conditions such as coronary artery disease, breast cancer, type I and II diabetes, individuals with outlier polygenic scores (e.g., top few percent) have been shown to have 5 or even 10 times higher risk than average. Several psychiatric conditions such as schizophrenia and autism also fall into this category. We discuss related topics such as the genetic architecture of complex traits, sibling validation of polygenic scores, and applications to adult health, in vitro fertilization (embryo selection), and genetic engineering.

Genome-Wide Association Study Statistical Models: A Review

Statistical models are at the core of the genome-wide association study (GWAS). In this chapter, we provide an overview of single- and multilocus statistical models, Bayesian, and machine learning approaches for association studies in plants. These models are discussed based on their basic methodology, cofactors adjustment accounted for, statistical power and computational efficiency. New statistical models and machine learning algorithms are both showing improved performance in detecting missed signals, rare mutations and prioritizing causal genetic variants; nevertheless, further optimization and validation studies are required to maximize the power of GWAS.

Construction and Clinical Translation of Causal Pan-Cancer Gene Score Across Cancer Types

Pan-cancer strategy, an integrative analysis of different cancer types, can be used to explain oncogenesis and identify biomarkers using a larger statistical power and robustness. Fine-mapping defines the casual loci, whereas genome-wide association studies (GWASs) typically identify thousands of cancer-related loci and not necessarily have a fine-mapping component. In this study, we develop a novel strategy to identify the causal loci using a pan-cancer and fine-mapping assumption, constructing the CAusal Pan-cancER gene (CAPER) score and validating its performance using internal and external validation on 1,287 individuals and 985 cell lines. Summary statistics of 15 cancer types were used to define 54 causal loci in 15 potential genes. Using the Cancer Genome Atlas (TCGA) training set, we constructed the CAPER score and divided cancer patients into two groups. Using the three validation sets, we found that 19 cancer-related variables were statistically significant between the two CAPER score groups and that 81 drugs had significantly different drug sensitivity between the two CAPER score groups. We hope that our strategies for selecting causal genes and for constructing CAPER score would provide valuable clues for guiding the management of different types of cancers.

MicroRNA Variants and HLA-miRNA Interactions are Novel Rheumatoid Arthritis Susceptibility Factors

Genome-wide association studies have identified >100 genetic risk factors for rheumatoid arthritis. However, the reported genetic variants could only explain less than 40% heritability of rheumatoid arthritis. The majority of the heritability is still missing and needs to be identified with more studies with different approaches and populations. In order to identify novel function SNPs to explain missing heritability and reveal novel mechanism pathogenesis of rheumatoid arthritis, 4 HLA SNPs (HLA-DRB1, HLA-DRB9, HLA-DQB1, and TNFAIP3) and 225 common SNPs located in miRNA, which might influence the miRNA target binding or pre-miRNA stability, were genotyped in 1,607 rheumatoid arthritis and 1,580 matched normal individuals. We identified 2 novel SNPs as significantly associated with rheumatoid arthritis including rs1414273 (miR-548ac, OR = 0.84, p = 8.26 × 10^-4) and rs2620381 (miR-627, OR = 0.77, p = 2.55 × 10^-3). We also identified that rs5997893 (miR-3928) showed significant epistasis effect with rs4947332 (HLA-DRB1, OR = 4.23, p = 0.04) and rs2967897 (miR-5695) with rs7752903 (TNFAIP3, OR = 4.43, p = 0.03). In addition, we found that individuals who carried 8 risk alleles showed 15.38 (95%CI: 4.69-50.49, p < 1.0 × 10^-6) times more risk of being affected by RA. Finally, we demonstrated that the targets of the significant miRNAs showed enrichment in immune related genes (p = 2.0 × 10^-5) and FDA approved drug target genes (p = 0.014). Overall, 6 novel miRNA SNPs including rs1414273 (miR-548ac, p = 8.26 × 10^-4), rs2620381 (miR-627, p = 2.55 × 10^-3), rs4285314 (miR-3135b, p = 1.10 × 10^-13), rs28477407 (miR-4308, p = 3.44 × 10^-5), rs5997893 (miR-3928, p = 5.9 × 10^-3) and rs45596840 (miR-4482, p = 6.6 × 10^-3) were confirmed to be significantly associated with RA in a Chinese population. Our study suggests that miRNAs might be interesting targets to accelerate understanding of the pathogenesis and drug development for rheumatoid arthritis.

Fast and Scalable Private Genotype Imputation Using Machine Learning and Partially Homomorphic Encryption

The recent advances in genome sequencing technologies provide unprecedented opportunities to understand the relationship between human genetic variation and diseases. However, genotyping whole genomes from a large cohort of individuals is still cost prohibitive. Imputation methods to predict genotypes of missing genetic variants are widely used, especially for genome-wide association studies. Accurate genotype imputation requires complex statistical methods. Due to the data and computing-intensive nature of the problem, imputation is increasingly outsourced, raising serious privacy concerns. In this work, we investigate solutions for fast, scalable, and accurate privacy-preserving genotype imputation using Machine Learning (ML) and a standardized homomorphic encryption scheme, Paillier cryptosystem. ML-based privacy-preserving inference has been largely optimized for computation-heavy non-linear functions in a single-output multi-class classification setting. However, having a large number of multi-class outputs per genome per individual calls for further optimizations and/or approximations specific to this application. Here we explore the effectiveness of linear models for genotype imputation to convert them to privacy-preserving equivalents using standardized homomorphic encryption schemes. Our results show that performance of our privacy-preserving genotype imputation method is equivalent to the state-of-the-art plaintext solutions, achieving up to 99% micro area under curve score, even on real-world large-scale datasets up to 80,000 targets.

Statistical Learning Methods Applicable to Genome-Wide Association Studies on Unbalanced Case-Control Disease Data

Despite the fact that imbalance between case and control groups is prevalent in genome-wide association studies (GWAS), it is often overlooked. This imbalance is getting more significant and urgent as the rapid growth of biobanks and electronic health records have enabled the collection of thousands of phenotypes from large cohorts, in particular for diseases with low prevalence. The unbalanced binary traits pose serious challenges to traditional statistical methods in terms of both genomic selection and disease prediction. For example, the well-established linear mixed models (LMM) yield inflated type I error rates in the presence of unbalanced case-control ratios. In this article, we review multiple statistical approaches that have been developed to overcome the inaccuracy caused by the unbalanced case-control ratio, with the advantages and limitations of each approach commented. In addition, we also explore the potential for applying several powerful and popular state-of-the-art machine-learning approaches, which have not been applied to the GWAS field yet. This review paves the way for better analysis and understanding of the unbalanced case-control disease data in GWAS.

Tejaas: reverse regression increases power for detecting trans-eQTLs

Trans-acting expression quantitative trait loci (trans-eQTLs) account for ≥70% expression heritability and could therefore facilitate uncovering mechanisms underlying the origination of complex diseases. Identifying trans-eQTLs is challenging because of small effect sizes, tissue specificity, and a severe multiple-testing burden. Tejaas predicts trans-eQTLs by performing L2-regularized “reverse” multiple regression of each SNP on all genes, aggregating evidence from many small trans-effects while being unaffected by the strong expression correlations. Combined with a novel unsupervised k-nearest neighbor method to remove confounders, Tejaas predicts 18851 unique trans-eQTLs across 49 tissues from GTEx. They are enriched in open chromatin, enhancers, and other regulatory regions. Many overlap with disease-associated SNPs, pointing to tissue-specific transcriptional regulation mechanisms.

Revisiting genome-wide association studies from statistical modelling to machine learning

Over the last decade, genome-wide association studies (GWAS) have discovered thousands of genetic variants underlying complex human diseases and agriculturally important traits. These findings have been utilized to dissect the biological basis of diseases, to develop new drugs, to advance precision medicine and to boost breeding. However, the potential of GWAS is still underexploited due to methodological limitations. Many challenges have emerged, including detecting epistasis and single-nucleotide polymorphisms (SNPs) with small effects and distinguishing causal variants from other SNPs associated through linkage disequilibrium. These issues have motivated advancements in GWAS analyses in two contrasting cultures-statistical modelling and machine learning. In this review, we systematically present the basic concepts and the benefits and limitations in both methods. We further discuss recent efforts to mitigate their weaknesses. Additionally, we summarize the state-of-the-art tools for detecting the missed signals, ultrarare mutations and gene-gene interactions and for prioritizing SNPs. Our work can offer both theoretical and practical guidelines for performing GWAS analyses and for developing further new robust methods to fully exploit the potential of GWAS.

Adaptive weighted sum tests via LASSO method in multi-locus family-based association analysis

Family based multi-locus tests integrate information from individual loci by weighted averaging of the marginal statistics, and have been proven to be more efficient and robust than the single-locus tests in genetic association studies. The power depends on how much information the weights can extract from data. The currently published weighted sum methods are only applicable to either common or rare variants and may suffer from substantial power loss especially for rare variants. In this paper, we propose a novel data-driven weight to improve the power under both common and rare variant circumstances. We use the l₁ regularization in Least Absolute Shrinkage and Selection Operator (LASSO) regression to construct the weight serving as a simultaneously adaptive marker selection process. Simulations for a dichotomous phenotype demonstrated that our LASSO-based approach outperformed the existing multi-locus methods in the sense of providing the highest statistical power while well controlled type I error rate under different scenarios. We also applied our methods to a real dataset for rheumatoid arthritis (GAW15 Problem 2). Two groups of alleles, in which individual SNPs had only modest and non-significant effects, were detected (P < 0.00001) using our proposed methods, whereas traditional multi-locus methods failed to identify them. In conclusion, the novel LASSO-based approach represents a superior weight-choosing strategy for multi-locus tests.

Leveraging mouse chromatin data for heritability enrichment informs common disease architecture and reveals cortical layer contributions to schizophrenia

Genome-wide association studies have implicated thousands of noncoding variants across common human phenotypes. However, they cannot directly inform the cellular context in which disease-associated variants act. Here, we use open chromatin profiles from discrete mouse cell populations to address this challenge. We applied stratified linkage disequilibrium score regression and evaluated heritability enrichment in 64 genome-wide association studies, emphasizing schizophrenia. We provide evidence that mouse-derived human open chromatin profiles can serve as powerful proxies for difficult to obtain human cell populations, facilitating the illumination of common disease heritability enrichment across an array of human phenotypes. We demonstrate that signatures from discrete subpopulations of cortical excitatory and inhibitory neurons are significantly enriched for schizophrenia heritability with maximal enrichment in cortical layer V excitatory neurons. We also show that differences between schizophrenia and bipolar disorder are concentrated in excitatory neurons in cortical layers II-III, IV, and V, as well as the dentate gyrus. Finally, we leverage these data to fine-map variants in 177 schizophrenia loci nominating variants in 104/177. We integrate these data with transcription factor binding site, chromatin interaction, and validated enhancer data, placing variants in the cellular context where they may modulate risk.

Evaluating the Potential of Younger Cases and Older Controls Cohorts to Improve Discovery Power in Genome-Wide Association Studies of Late-Onset Diseases

For more than a decade, genome-wide association studies have been making steady progress in discovering the causal gene variants that contribute to late-onset human diseases. Polygenic late-onset diseases in an aging population display a risk allele frequency decrease at older ages, caused by individuals with higher polygenic risk scores becoming ill proportionately earlier and bringing about a change in the distribution of risk alleles between new cases and the as-yet-unaffected population. This phenomenon is most prominent for diseases characterized by high cumulative incidence and high heritability, examples of which include Alzheimer’s disease, coronary artery disease, cerebral stroke, and type 2 diabetes, while for late-onset diseases with relatively lower prevalence and heritability, exemplified by cancers, the effect is significantly lower. In this research, computer simulations have demonstrated that genome-wide association studies of late-onset polygenic diseases showing high cumulative incidence together with high initial heritability will benefit from using the youngest possible age-matched cohorts. Moreover, rather than using age-matched cohorts, study cohorts combining the youngest possible cases with the oldest possible controls may significantly improve the discovery power of genome-wide association studies.