Search

OBJECTIVES: Population-level measures of seropositivity are critical for understanding the epidemiology of an emerging pathogen, yet most antibody tests apply a strict cutoff for seropositivity that is not learnt in a data-driven manner, leading to uncertainty when classifying low-titer responses. To improve upon this, we evaluated cutoff-independent methods for their ability to assign likelihood of SARS-CoV-2 seropositivity to individual samples. METHODS: Using robust ELISAs based on SARS-CoV-2 spike (S) and the receptor-binding domain (RBD), we profiled antibody responses in a group of SARS-CoV-2 PCR+ individuals (n = 138). Using these data, we trained probabilistic learners to assign likelihood of seropositivity to test samples of unknown serostatus (n = 5100), identifying a support vector machines-linear discriminant analysis learner (SVM-LDA) suited for this purpose. RESULTS: In the training data from confirmed ancestral SARS-CoV-2 infections, 99% of participants had detectable anti-S and -RBD IgG in the circulation, with titers differing > 1000-fold between persons. In data of otherwise healthy individuals, 7.2% (n = 367) of samples were of uncertain serostatus, with values in the range of 3-6SD from the mean of pre-pandemic negative controls (n = 595). In contrast, SVM-LDA classified 6.4% (n = 328) of test samples as having a high likelihood (> 99% chance) of past infection, 4.5% (n = 230) to have a 50-99% likelihood, and 4.0% (n = 203) to have a 10-49% likelihood. As different probabilistic approaches were more consistent with each other than conventional SD-based methods, such tools allow for more statistically-sound seropositivity estimates in large cohorts. CONCLUSION: Probabilistic antibody testing frameworks can improve seropositivity estimates in populations with large titer variability.

Many genetic mutations affecting phenotypes are presumed to do so via altering gene expression in particular cells or tissues, but identifying the specific genes involved has been challenging. A transcriptome-wide association study (TWAS) attempts to identify disease associated genes by first learning a predictive model on an eQTL dataset and then imputing gene expression levels into a larger genome-wide association study (GWAS). Finally, associations between predicted gene expressions and GWAS phenotype are identified.Here, we compared tree-based machine learning (ML) method of random forests (RF) with more widely used linear methods of lasso, ridge, and elastic net regression, for prediction of gene expression. We also developed a multi-task learning extension to RF which simultaneously makes use of information from multiple tissues (RF-MTL) and compared it to a multi-dataset version of lasso, the joint lasso, and to a single tissue RF. We found that for prediction of gene expression, RF, in general, outperformed linear approaches on our chosen eQTL dataset and that multi-tissue methods generally outperformed their single-tissue counterparts, with RF-MTL performing the best. Simulations showed that these benefits generally propagated to the next steps of the analysis, although highlighted that joint lasso had a tendency to erroneously identify genes in one tissue if there existed a disease signal for that gene in another.We tested all four methods on type 1 diabetes (T1D) GWAS and expression data for several immune cells and found that 46 genes were identified by at least one method, though only 7 by all methods. Joint lasso discovered the most T1D-associated genes, including 15 unique to that method, but this may reflect its higher false positive rate due to “overborrowing” information across tissues. RF-MTL found more unique associated genes than RF for 3 out 5 tissues. Compared to lasso-based analysis, the RF gene list was more likely to relate to T1D in an analysis of independent data types. We conclude that RF, both single- and multi-task version, is competitive and, for some cell types, superior to linear models conventionally used in the TWAS studies.Author summaryA transcriptome-wide association study (TWAS) is a way of integrating expression data and genome-wide association studies (GWAS), which allows for discovery of genes, rather than mutations, associated to traits of interest. In the TWAS framework, we first train predictive models on an eQTL dataset, then use these models to impute gene expression into a GWAS dataset. Finally, we look for significant associations between predicted gene expression and a GWAS trait. In this work, we compare non-linear method of random forests (RF) to linear models, customarily used in TWAS. Furthermore, we demonstrate that TWAS framework can naturally be extended to, and potentially benefit from, a multi-tissue setting, thereby taking advantage of the correlation between gene expression in different tissue types. We applied the RF, a selection of linear models, and the multi-tissue approaches to an eQTL dataset of monocytes and B cells and a large T1D GWAS. We found that RF outperform lasso in terms of predictive accuracy and the number of differentially expressed genes found, and that multi-dataset version of lasso discovered the most T1D-associated genes. Analysis of the gene lists produced for each method in independent data types (excluding genetic association data) showed all related to T1D, but that the RF methods ranked T1D higher in their lists than the linear methods. We conclude that RF is a useful addition to the TWAS tool box.

In phenotype prediction the physical characteristics of an organism are predicted from knowledge of its genotype and environment. Such studies, often called genome-wide association studies, are of the highest societal importance, as they are of central importance to medicine, crop-breeding, etc. We investigated three phenotype prediction problems: one simple and clean (yeast), and the other two complex and real-world (rice and wheat). We compared standard machine learning methods; elastic net, ridge regression, lasso regression, random forest, gradient boosting machines (GBM), and support vector machines (SVM), with two state-of-the-art classical statistical genetics methods; genomic BLUP and a two-step sequential method based on linear regression. Additionally, using the clean yeast data, we investigated how performance varied with the complexity of the biological mechanism, the amount of observational noise, the number of examples, the amount of missing data, and the use of different data representations. We found that for almost all the phenotypes considered, standard machine learning methods outperformed the methods from classical statistical genetics. On the yeast problem, the most successful method was GBM, followed by lasso regression, and the two statistical genetics methods; with greater mechanistic complexity GBM was best, while in simpler cases lasso was superior. In the wheat and rice studies the best two methods were SVM and BLUP. The most robust method in the presence of noise, missing data, etc. was random forests. The classical statistical genetics method of genomic BLUP was found to perform well on problems where there was population structure. This suggests that standard machine learning methods need to be refined to include population structure information when this is present. We conclude that the application of machine learning methods to phenotype prediction problems holds great promise, but that determining which methods is likely to perform well on any given problem is elusive and non-trivial.

Cytokines are essential regulatory components of the immune system, and their aberrant levels have been linked to many disease states. Despite increasing evidence that cytokines operate in concert, many of the physiological interactions between cytokines, and the shared genetic architecture that underlies them, remain unknown. Here, we aimed to identify and characterize genetic variants with pleiotropic effects on cytokines. Using three population-based cohorts (n = 9,263), we performed multivariate genome-wide association studies (GWAS) for a correlation network of 11 circulating cytokines, then combined our results in meta-analysis. We identified a total of eight loci significantly associated with the cytokine network, of which two (PDGFRB and ABO) had not been detected previously. In addition, conditional analyses revealed a further four secondary signals at three known cytokine loci. Integration, through the use of Bayesian colocalization analysis, of publicly available GWAS summary statistics with the cytokine network associations revealed shared causal variants between the eight cytokine loci and other traits; in particular, cytokine network variants at the ABO, SERPINE2, and ZFPM2 loci showed pleiotropic effects on the production of immune-related proteins, on metabolic traits such as lipoprotein and lipid levels, on blood-cell-related traits such as platelet count, and on disease traits such as coronary artery disease and type 2 diabetes.

Thousands of genetic variants are associated with human disease risk, but linkage disequilibrium (LD) hinders fine-mapping the causal variants. Both lack of power, and joint tagging of two or more distinct causal variants by a single non-causal SNP, lead to inaccuracies in fine-mapping, with stochastic search more robust than stepwise. We develop a computationally efficient multinomial fine-mapping (MFM) approach that borrows information between diseases in a Bayesian framework. We show that MFM has greater accuracy than single disease analysis when shared causal variants exist, and negligible loss of precision otherwise. MFM analysis of six immune-mediated diseases reveals causal variants undetected in individual disease analysis, including in IL2RA where we confirm functional effects of multiple causal variants using allele-specific expression in sorted CD4+ T cells from genotype-selected individuals. MFM has the potential to increase fine-mapping resolution in related diseases enabling the identification of associated cellular and molecular phenotypes.

Thousands of genetic variants are associated with human disease risk, but linkage disequilibrium (LD) hinders fine-mapping the causal variants. Both lack of power, and joint tagging of two or more distinct causal variants by a single non-causal SNP, lead to inaccuracies in fine-mapping, with stochastic search more robust than stepwise. We develop a computationally efficient multinomial fine-mapping (MFM) approach that borrows information between diseases in a Bayesian framework. We show that MFM has greater accuracy than single disease analysis when shared causal variants exist, and negligible loss of precision otherwise. MFM analysis of six immune-mediated diseases reveals causal variants undetected in individual disease analysis, including in IL2RA where we confirm functional effects of multiple causal variants using allele-specific expression in sorted CD4+ T cells from genotype-selected individuals. MFM has the potential to increase fine-mapping resolution in related diseases enabling the identification of associated cellular and molecular phenotypes.

In phenotype prediction the physical characteristics of an organism are predicted from knowledge of its genotype and environment. Such studies, often called genome-wide association studies, are of the highest societal importance, as they are of central importance to medicine, crop-breeding, etc. We investigated three phenotype prediction problems: one simple and clean (yeast), and the other two complex and real-world (rice and wheat). We compared standard machine learning methods; elastic net, ridge regression, lasso regression, random forest, gradient boosting machines (GBM), and support vector machines (SVM), with two state-of-the-art classical statistical genetics methods; genomic BLUP and a two-step sequential method based on linear regression. Additionally, using the clean yeast data, we investigated how performance varied with the complexity of the biological mechanism, the amount of observational noise, the number of examples, the amount of missing data, and the use of different data representations. We found that for almost all the phenotypes considered, standard machine learning methods outperformed the methods from classical statistical genetics. On the yeast problem, the most successful method was GBM, followed by lasso regression, and the two statistical genetics methods; with greater mechanistic complexity GBM was best, while in simpler cases lasso was superior. In the wheat and rice studies the best two methods were SVM and BLUP. The most robust method in the presence of noise, missing data, etc. was random forests. The classical statistical genetics method of genomic BLUP was found to perform well on problems where there was population structure. This suggests that standard machine learning methods need to be refined to include population structure information when this is present. We conclude that the application of machine learning methods to phenotype prediction problems holds great promise, but that determining which methods is likely to perform well on any given problem is elusive and non-trivial.

Thousands of genetic variants are associated with human disease risk, but linkage disequilibrium (LD) hinders fine-mapping the causal variants. Both lack of power, and joint tagging of two or more distinct causal variants by a single non-causal SNP, lead to inaccuracies in fine-mapping, with stochastic search more robust than stepwise. We develop a computationally efficient multinomial fine-mapping (MFM) approach that borrows information between diseases in a Bayesian framework. We show that MFM has greater accuracy than single disease analysis when shared causal variants exist, and negligible loss of precision otherwise. MFM analysis of six immune-mediated diseases reveals causal variants undetected in individual disease analysis, including in IL2RA where we confirm functional effects of multiple causal variants using allele-specific expression in sorted CD4+ T cells from genotype-selected individuals. MFM has the potential to increase fine-mapping resolution in related diseases enabling the identification of associated cellular and molecular phenotypes., Statistical fine-mapping to pinpoint likely causal variants in a genomic region is complicated by linkage disequilibrium (LD). Here, Asimit et al. compare stepwise and stochastic approaches to fine-mapping and propose a Bayesian multinomial stochastic search method which they apply to six immune-mediated diseases.