arXiv subjects

Tom Michoel

Publications and source records attributed to Tom Michoel.

At least 19 recordsLinked to original sources

arXiv2025

A network-driven framework for enhancing gene-disease association studies in coronary artery disease

Transcriptome-wide association studies (TWAS) link genetic variation to complex traits by leveraging expression quantitative trait loci (eQTL) data. However, most implementations are typically limited to local (cis-acting) effects and fail to account for long-range (trans) regulatory influences mediated through gene networks. We introduce GRN-TWAS, a framework that reconstructs gene regulatory networks (GRNs) and integrates their topology into gene expression prediction models, thereby propagating distal (trans) regulatory effects through tissue-specific gene networks to trait- or disease-associated phenotypes. By incorporating network-derived trans-eQTLs, GRN-TWAS generates gene expression imputation models that capture both local and distal genetic components, enabling a more complete, systems-level view of genetic regulation consistent with the omnigenic model hypothesis. Using genotype and multi-tissue expression data from 600 coronary artery disease (CAD) cases in the STARNET study together with GWAS summary statistics, we show that GRN-TWAS improves gene-expression prediction and sharpens discovery of CAD-associated genes. Across seven tissues, the framework identified 5,779 transcriptome-wide significant genes, more than 50\% of which appear to be previously unreported in the CAD literature. A knowledge-based gene-ranking engine then prioritized 882 genes as highly CAD-relevant, including 237 regulated exclusively through trans effects. Key-driver analysis highlighted 18 putative trans mediators with high network centrality and disease relevance, offering mechanistic hypotheses that complement association signals. Collectively, these results demonstrate that embedding network topology into TWAS improves discovery and interpretability by exposing tissue-specific regulatory routes from genotype to phenotype and expanding the landscape of gene-disease associations.

q-bio.MN↗

arXiv2024

Predicting the genetic component of gene expression using gene regulatory networks

Gene expression prediction plays a vital role in transcriptome-wide association studies (TWAS), which seek to establish associations between tissue gene expression and complex traits. Traditional models rely on genetic variants in close genomic proximity to the gene of interest to predict the genetic component of gene expression. In this study, we propose a novel approach incorporating distal genetic variants acting through gene regulatory networks (GRNs) into gene expression prediction models, in line with the omnigenic model of complex trait inheritance. Using causal and coexpression GRNs reconstructed from genomic and transcriptomic data and modeling the data as a Bayesian network jointly over genetic variants and genes, inference of gene expression from observed genotypic data is achieved through a two-step process. Initially, the expression level of each gene in the network is predicted using its local genetic variants. The residuals, calculated as the differences between the observed and predicted expression levels, are then modeled using the genotype information of parent and/or grandparent nodes in the GRN. The final predicted expression level of the gene is obtained by summing the predictions from the local variants model and the residual model, effectively incorporating both local and distal genetic influences. Using various regularized regression techniques for parameter estimation, we found that GRN-based gene expression prediction outperformed the traditional local-variant approach on simulated data from the DREAM5 Systems Genetics Challenge and real data from the Geuvadis study and an eQTL mapping study in yeast. This study provides important insights into the challenge of gene expression prediction for TWAS. It reaffirms the importance of GRNs for understanding the genetic effects on gene expression and complex traits more generally.

q-bio.MN↗

arXiv2024

Prediction of causal genes at GWAS loci with pleiotropic gene regulatory effects using sets of correlated instrumental variables

Multivariate Mendelian randomization (MVMR) is a statistical technique that uses sets of genetic instruments to estimate the direct causal effects of multiple exposures on an outcome of interest. At genomic loci with pleiotropic gene regulatory effects, that is, loci where the same genetic variants are associated to multiple nearby genes, MVMR can potentially be used to predict candidate causal genes. However, consensus in the field dictates that the genetic instruments in MVMR must be independent, which is usually not possible when considering a group of candidate genes from the same locus. We used causal inference theory to show that MVMR with correlated instruments satisfies the instrumental set condition. This is a classical result by Brito and Pearl (2002) for structural equation models that guarantees the identifiability of causal effects in situations where multiple exposures collectively, but not individually, separate a set of instrumental variables from an outcome variable. Extensive simulations confirmed the validity and usefulness of these theoretical results even at modest sample sizes. Importantly, the causal effect estimates remain unbiased and their variance small when instruments are highly correlated. We applied MVMR with correlated instrumental variable sets at risk loci from genome-wide association studies (GWAS) for coronary artery disease using eQTL data from the STARNET study. Our method predicts causal genes at twelve loci, each associated with multiple colocated genes in multiple tissues. However, the extensive degree of regulatory pleiotropy across tissues and the limited number of causal variants in each locus still require that MVMR is run on a tissue-by-tissue basis, and testing all gene-tissue pairs at a given locus in a single model to predict causal gene-tissue combinations remains infeasible.

stat.ME↗

arXiv2022

Causal inference in drug discovery and development

To discover new drugs is to seek and to prove causality. As an emerging approach leveraging human knowledge and creativity, data, and machine intelligence, causal inference holds the promise of reducing cognitive bias and improving decision making in drug discovery. While it has been applied across the value chain, the concepts and practice of causal inference remain obscure to many practitioners. This article offers a non-technical introduction to causal inference, reviews its recent applications, and discusses opportunities and challenges of adopting the causal language in drug discovery and development.

q-bio.QM↗

arXiv2022

rfPhen2Gen: A machine learning based association study of brain imaging phenotypes to genotypes

Imaging genetic studies aim to find associations between genetic variants and imaging quantitative traits. Traditional genome-wide association studies (GWAS) are based on univariate statistical tests, but when multiple traits are analyzed together they suffer from a multiple-testing problem and from not taking into account correlations among the traits. An alternative approach to multi-trait GWAS is to reverse the functional relation between genotypes and traits, by fitting a multivariate regression model to predict genotypes from multiple traits simultaneously. However, current reverse genotype prediction approaches are mostly based on linear models. Here, we evaluated random forest regression (RFR) as a method to predict SNPs from imaging QTs and identify biologically relevant associations. We learned machine learning models to predict 518,484 SNPs using 56 brain imaging QTs. We observed that genotype regression error is a better indicator of permutation p-value significance than genotype classification accuracy. SNPs within the known Alzheimer disease (AD) risk gene APOE had lowest RMSE for lasso and random forest, but not ridge regression. Moreover, random forests identified additional SNPs that were not prioritized by the linear models but are known to be associated with brain-related disorders. Feature selection identified well-known brain regions associated with AD,like the hippocampus and amygdala, as important predictors of the most significant SNPs. In summary, our results indicate that non-linear methods like random forests may offer additional insights into phenotype-genotype associations compared to traditional linear multi-variate GWAS methods.

q-bio.GN↗

arXiv2021

High-dimensional multi-trait GWAS by reverse prediction of genotypes

Multi-trait genome-wide association studies (GWAS) use multi-variate statistical methods to identify associations between genetic variants and multiple correlated traits simultaneously, and have higher statistical power than independent univariate analyses of traits. Reverse regression, where genotypes of genetic variants are regressed on multiple traits simultaneously, has emerged as a promising approach to perform multi-trait GWAS in high-dimensional settings where the number of traits exceeds the number of samples. We analyzed different machine learning methods (ridge regression, naive Bayes/independent univariate, random forests and support vector machines) for reverse regression in multi-trait GWAS, using genotypes, gene expression data and ground-truth transcriptional regulatory networks from the DREAM5 SysGen Challenge and from a cross between two yeast strains to evaluate methods. We found that genotype prediction performance, in terms of root mean squared error (RMSE), allowed to distinguish between genomic regions with high and low transcriptional activity. Moreover, model feature coefficients correlated with the strength of association between variants and individual traits, and were predictive of true trans-eQTL target genes, with complementary findings across methods. Code to reproduce the analysis is available at https://github.com/michoel-lab/Reverse-Pred-GWAS

q-bio.GN↗

arXiv2021

Integrating Sensing and Communication in Cellular Networks via NR Sidelink

RF-sensing, the analysis and interpretation of movement or environment-induced patterns in received electromagnetic signals, has been actively investigated for more than a decade. Since electromagnetic signals, through cellular communication systems, are omnipresent, RF sensing has the potential to become a universal sensing mechanism with applications in smart home, retail, localization, gesture recognition, intrusion detection, etc. Specifically, existing cellular network installations might be dual-used for both communication and sensing. Such communications and sensing convergence is envisioned for future communication networks. We propose the use of NR-sidelink direct device-to-device communication to achieve device-initiated,flexible sensing capabilities in beyond 5G cellular communication systems. In this article, we specifically investigate a common issue related to sidelink-based RF-sensing, which is its angle and rotation dependence. In particular, we discuss transformations of mmWave point-cloud data which achieve rotational invariance, as well as distributed processing based on such rotational invariant inputs, at angle and distance diverse devices. To process the distributed data, we propose a graph based encoder to capture spatio-temporal features of the data and propose four approaches for multi-angle learning. The approaches are compared on a newly recorded and openly available dataset comprising 15 subjects, performing 21 gestures which are recorded from 8 angles.

cs.CV↗

arXiv2021

Tesla-Rapture: A Lightweight Gesture Recognition System from mmWave Radar Point Clouds

We present Tesla-Rapture, a gesture recognition interface for point clouds generated by mmWave Radars. State of the art gesture recognition models are either too resource consuming or not sufficiently accurate for integration into real-life scenarios using wearable or constrained equipment such as IoT devices (e.g. Raspberry PI), XR hardware (e.g. HoloLens), or smart-phones. To tackle this issue, we developed Tesla, a Message Passing Neural Network (MPNN) graph convolution approach for mmWave radar point clouds. The model outperforms the state of the art on two datasets in terms of accuracy while reducing the computational complexity and, hence, the execution time. In particular, the approach, is able to predict a gesture almost 8 times faster than the most accurate competitor. Our performance evaluation in different scenarios (environments, angles, distances) shows that Tesla generalizes well and improves the accuracy up to 20% in challenging scenarios like a through-wall setting and sensing at extreme angles. Utilizing Tesla, we develop Tesla-Rapture, a real-time implementation using a mmWave Radar on a Raspberry PI 4 and evaluate its accuracy and time-complexity. We also publish the source code, the trained models, and the implementation of the model for embedded devices.

cs.CV↗

arXiv2020

A Graph Feature Auto-Encoder for the Prediction of Unobserved Node Features on Biological Networks

Motivation: Molecular interaction networks summarize complex biological processes as graphs, whose structure is informative of biological function at multiple scales. Simultaneously, omics technologies measure the variation or activity of genes, proteins, or metabolites across individuals or experimental conditions. Integrating the complementary viewpoints of biological networks and omics data is an important task in bioinformatics, but existing methods treat networks as discrete structures, which are intrinsically difficult to integrate with continuous node features or activity measures. Graph neural networks map graph nodes into a low-dimensional vector space representation, and can be trained to preserve both the local graph structure and the similarity between node features. Results: We studied the representation of transcriptional, protein-protein and genetic interaction networks in E. Coli and mouse using graph neural networks. We found that such representations explain a large proportion of variation in gene expression data, and that using gene expression data as node features improves the reconstruction of the graph from the embedding. We further proposed a new end-to-end graph feature auto-encoder which is trained on the feature reconstruction task, and showed that it performs better at predicting unobserved node features than auto-encoders that are trained on the graph reconstruction task before learning to predict node features. When applied to the problem of imputing missing data in single-cell RNAseq data, our graph feature auto-encoder outperformed a state-of-the-art imputation method that does not use protein interaction information, showing the benefit of integrating biological networks and omics data using graph representation learning.

q-bio.QM↗

arXiv2020

Comparison between instrumental variable and mediation-based methods for reconstructing causal gene networks in yeast

Causal gene networks model the flow of information within a cell, but reconstructing them from omics data is challenging because correlation does not imply causation. Combining genomics and transcriptomics data from a segregating population allows to orient the direction of causality between gene expression traits using genomic variants. Instrumental-variable methods (IV) use a local expression quantitative trait locus (eQTL) as a randomized instrument for a gene's expression level, and assign target genes based on distal eQTL associations. Mediation-based methods (ME) additionally require that distal eQTL associations are mediated by the source gene. Here we used Findr, a software providing uniform implementations of IV, ME, and coexpression-based methods, a recent dataset of 1,012 segregants from a cross between two budding yeast strains, and the YEASTRACT database of known transcriptional interactions to compare causal gene network inference methods. We found that causal inference methods result in a significant overlap with the ground-truth, whereas coexpression did not perform better than random. A subsampling analysis revealed that the performance of ME decreases at large sample sizes, due to a loss of sensitivity when residual correlations become significant. IV methods contain false positive predictions, due to genomic linkage between eQTL instruments. IV and ME methods also have complementary roles for identifying causal genes underlying transcriptional hotspots. IV methods correctly predicted STB5 targets for a hotspot centred on the transcription factor STB5, whereas ME failed due to Stb5p auto-regulating its own expression. ME suggests a new candidate gene, DNM1, for a hotspot on Chr XII, where IV methods could not distinguish between multiple genes located within the hotspot.

q-bio.MN↗

arXiv2020

Restricted maximum-likelihood method for learning latent variance components in gene expression data with known and unknown confounders

Random effect models are popular statistical models for detecting and correcting spurious sample correlations due to hidden confounders in genome-wide gene expression data. In applications where some confounding factors are known, estimating simultaneously the contribution of known and latent variance components in random effect models is a challenge that has so far relied on numerical gradient-based optimizers to maximize the likelihood function. This is unsatisfactory because the resulting solution is poorly characterized and the efficiency of the method may be suboptimal. Here we prove analytically that maximum-likelihood latent variables can always be chosen orthogonal to the known confounding factors, in other words, that maximum-likelihood latent variables explain sample covariances not already explained by known factors. Based on this result we propose a restricted maximum-likelihood method which estimates the latent variables by maximizing the likelihood on the restricted subspace orthogonal to the known confounding factors, and show that this reduces to probabilistic PCA on that subspace. The method then estimates the variance-covariance parameters by maximizing the remaining terms in the likelihood function given the latent variables, using a newly derived analytic solution for this problem. Compared to gradient-based optimizers, our method attains greater or equal likelihood values, can be computed using standard matrix operations, results in latent factors that don't overlap with any known factors, and has a runtime reduced by several orders of magnitude. Hence the restricted maximum-likelihood method facilitates the application of random effect modelling strategies for learning latent variance components to much larger gene expression datasets than possible with current methods.

stat.ME↗

arXiv2018

Model-based clustering of multi-tissue gene expression data

Recently, it has become feasible to generate large-scale, multi-tissue gene expression data, where expression profiles are obtained from multiple tissues or organs sampled from dozens to hundreds of individuals. When traditional clustering methods are applied to this type of data, important information is lost, because they either require all tissues to be analyzed independently, ignoring dependencies and similarities between tissues, or to merge tissues in a single, monolithic dataset, ignoring individual characteristics of tissues. We developed a Bayesian model-based multi-tissue clustering algorithm, revamp, which can incorporate prior information on physiological tissue similarity, and which results in a set of clusters, each consisting of a core set of genes conserved across tissues as well as differential sets of genes specific to one or more subsets of tissues. Using data from seven vascular and metabolic tissues from over 100 individuals in the STockholm Atherosclerosis Gene Expression (STAGE) study, we demonstrate that multi-tissue clusters inferred by revamp are more enriched for tissue-dependent protein-protein interactions compared to alternative approaches. We further demonstrate that revamp results in easily interpretable multi-tissue gene expression associations to key coronary artery disease processes and clinical phenotypes in the STAGE individuals. Revamp is implemented in the Lemon-Tree software, available at https://github.com/eb00/lemon-tree

q-bio.QM↗

arXiv2018

Controlling false discoveries in Bayesian gene networks with lasso regression p-values

Bayesian networks can represent directed gene regulations and therefore are favored over co-expression networks. However, hardly any Bayesian network study concerns the false discovery control (FDC) of network edges, leading to low accuracies due to systematic biases from inconsistent false discovery levels in the same study. We design four empirical tests to examine the FDC of Bayesian networks from three p-value based lasso regression variable selections --- two existing and one we originate. Our method, lassopv, computes p-values for the critical regularization strength at which a predictor starts to contribute to lasso regression. Using null and Geuvadis datasets, we find that lassopv obtains optimal FDC in Bayesian gene networks, whilst existing methods have defective p-values. The FDC concept and tests extend to most network inference scenarios and will guide the design and improvement of new and existing methods. Our novel variable selection method with lasso regression also allows FDC on other datasets and questions, even beyond network inference and computational biology. Lassopv is implemented in R and freely available at https://github.com/lingfeiwang/lassopv and https://cran.r-project.org/package=lassopv

stat.ME↗

arXiv2018

Learning differential module networks across multiple experimental conditions

Module network inference is a statistical method to reconstruct gene regulatory networks, which uses probabilistic graphical models to learn modules of coregulated genes and their upstream regulatory programs from genome-wide gene expression and other omics data. Here we review the basic theory of module network inference, present protocols for common gene regulatory network reconstruction scenarios based on the Lemon-Tree software, and show, using human gene expression data, how the software can also be applied to learn differential module networks across multiple experimental conditions.

q-bio.QM↗

arXiv2017

Wisdom of the crowd from unsupervised dimension reduction

Wisdom of the crowd, the collective intelligence derived from responses of multiple human or machine individuals to the same questions, can be more accurate than each individual, and improve social decision-making and prediction accuracy. This can also integrate multiple programs or datasets, each as an individual, for the same predictive questions. Crowd wisdom estimates each individual's independent error level arising from their limited knowledge, and finds the crowd consensus that minimizes the overall error. However, previous studies have merely built isolated, problem-specific models with limited generalizability, and mainly for binary (yes/no) responses. Here we show with simulation and real-world data that the crowd wisdom problem is analogous to one-dimensional unsupervised dimension reduction in machine learning. This provides a natural class of crowd wisdom solutions, such as principal component analysis and Isomap, which can handle binary and also continuous responses, like confidence levels, and consequently can be more accurate than existing solutions. They can even outperform supervised-learning-based collective intelligence that is calibrated on historical performance of individuals, e.g. penalized linear regression and random forest. This study unifies crowd wisdom and unsupervised dimension reduction, and thereupon introduces a broad range of highly-performing and widely-applicable crowd wisdom methods. As the costs for data acquisition and processing rapidly decrease, this study will promote and guide crowd wisdom applications in the social and natural sciences, including data fusion, meta-analysis, crowd-sourcing, and committee decision making.

stat.ML↗

arXiv2017

Analytic solution and stationary phase approximation for the Bayesian lasso and elastic net

The lasso and elastic net linear regression models impose a double-exponential prior distribution on the model parameters to achieve regression shrinkage and variable selection, allowing the inference of robust models from large data sets. However, there has been limited success in deriving estimates for the full posterior distribution of regression coefficients in these models, due to a need to evaluate analytically intractable partition function integrals. Here, the Fourier transform is used to express these integrals as complex-valued oscillatory integrals over "regression frequencies". This results in an analytic expansion and stationary phase approximation for the partition functions of the Bayesian lasso and elastic net, where the non-differentiability of the double-exponential prior has so far eluded such an approach. Use of this approximation leads to highly accurate numerical estimates for the expectation values and marginal posterior distributions of the regression coefficients, and allows for Bayesian inference of much higher dimensional models than previously possible.

stat.ME↗

arXiv2017

Efficient and accurate causal inference with hidden confounders from genome-transcriptome variation data

Mapping gene expression as a quantitative trait using whole genome-sequencing and transcriptome analysis allows to discover the functional consequences of genetic variation. We developed a novel method and ultra-fast software Findr for higly accurate causal inference between gene expression traits using cis-regulatory DNA variations as causal anchors, which improves current methods by taking into account hidden confounders and weak regulations. Findr outperformed existing methods on the DREAM5 Systems Genetics challenge and on the prediction of microRNA and transcription factor targets in human lymphoblastoid cells, while being nearly a million times faster. Findr is publicly available at https://github.com/lingfeiwang/findr

q-bio.GN↗

arXiv2016

Detection of regulator genes and eQTLs in gene networks

Genetic differences between individuals associated to quantitative phenotypic traits, including disease states, are usually found in non-coding genomic regions. These genetic variants are often also associated to differences in expression levels of nearby genes (they are "expression quantitative trait loci" or eQTLs for short) and presumably play a gene regulatory role, affecting the status of molecular networks of interacting genes, proteins and metabolites. Computational systems biology approaches to reconstruct causal gene networks from large-scale omics data have therefore become essential to understand the structure of networks controlled by eQTLs together with other regulatory genes, and to generate detailed hypotheses about the molecular mechanisms that lead from genotype to phenotype. Here we review the main analytical methods and softwares to identify eQTLs and their associated genes, to reconstruct co-expression networks and modules, to reconstruct causal Bayesian gene and module networks, and to validate predicted networks in silico.

q-bio.MN↗