Biology subjects

Lutz, S. M.

Publications and source records attributed to Lutz, S. M..

10 recordsLinked to original sources

bioRxiv2025

Analyzing associations and higher-order effects in multi-omics data with double machine learning

MotivationIntegrative omics analyses enhance our understanding of disease mechanisms and biomarkers by investigating relationships among traits, omics measurements, genetic variants, and epidemiological factors. Statistically, these analyses are challenging and require robust, flexible methodologies due to high dimensionality, non-standard data distributions, and potentially complex, non-linear confounding effects. ResultsTo facilitate the integration and analysis of multi-omics data, we introduce the Robust Omics MethodologY (ROMY) framework and its corresponding R implementation, the romy package. ROMY enables users to (a) perform robust association testing between two target variables while incorporating flexible covariate adjustments, (b) examine effects on measurement variances and covariances (e.g., co-expression, co-abundance), and (c) conduct rigorous interaction-effect testing. ROMY builds on recent advances in theoretical statistics and double machine learning to ensure robustness and statistical validity. We illustrate the performance of our framework through simulation studies. Availability and implementationromy is an R package available under the GNU GPLv3 license on GitHub: https://github.com/julianhecker/romy.

bioinformatics↗

bioRxiv2021

A robust and adaptive framework for interaction testing in quantitative traits between multiple genetic loci and exposure variables

The identification and understanding of gene-environment interactions can provide insights into the pathways and mechanisms underlying complex diseases. However, testing for gene-environment interaction remains a challenge since statistical power is often limited, the specification of environmental effects is nontrivial, and such misspecifications can lead to false positive findings. To address the lack of statistical power, recent methods aim to identify interactions on an aggregated level using, for example, polygenic risk scores. While this strategy increases power to detect interactions, identifying contributing key genes and pathways is difficult based on these global results. Here, we propose RITSS (Robust Interaction Testing using Sample Splitting), a gene-environment interaction testing framework for quantitative traits that is based on sample splitting and robust test statistics. RITSS can incorporate multiple genetic variants and/or multiple environmental factors. Using sample splitting, a screening step enables the selection and combination of potential interactions into scores with improved interpretability, based on the users unrestricted choices for statistical/machine learning approaches. In the testing step, the application of robust test statistics minimizes the susceptibility of the results to main effect misspecifications. Using extensive simulation studies, we demonstrate that RITSS controls the type 1 error rate in a wide range of scenarios. In an application to lung function phenotypes and human height in the UK Biobank, RITSS identified genome-wide significant interactions with subcomponents of genetic risk scores. While the contributing single variant interactions are moderate, our analysis results indicate interesting interaction patterns that result in strong aggregated signals that provide further insights into gene-environment interaction mechanisms.

bioinformatics↗

bioRxiv2021

Expanding the pool of public controls for GWAS via a method for combining genotypes from arrays and sequencing

Genome-wide association studies (GWAS) have made impactful discoveries for complex diseases, often by amassing very large sample sizes. Yet, GWAS of many diseases remain underpowered, especially for non-European ancestries. One cost-effective approach to increase sample size is to combine existing case-only cohorts with public controls, but this approach is limited by the need for a large overlap in variants across genotyping arrays and the scarcity of non-European controls. We developed and validated a protocol, Genotyping Array-WGS Merge (GAWMerge), for combining genotypes from arrays and whole genome sequencing, ensuring complete variant overlap, and allowing for diverse samples like Trans-Omics for Precision Medicine to be used. Our protocol involves phasing, imputation, and filtering. We illustrated its ability to control type I error and recover known disease-associated signals across technologies, independent datasets, and ancestries in smoking-related cohorts. GAWMerge enables genetic studies to leverage existing cohorts to validly increase sample size and enhance discovery.

bioinformatics↗

bioRxiv2021

A smoothed version of the Lassosum penalty for fitting integrated risk models

Polygenic risk scores are a popular means to predict the disease risk or disease susceptibility of an individual based on its genotype information. When adding other important epidemiological covariates such as age or sex, we speak of an integrated risk model. Methodological advances for fitting more accurate integrated risk models are of immediate importance to improve the precision of risk prediction, thereby potentially identifying patients at high risk early on when they are still able to benefit from preventive steps/interventions targeted at increasing their odds of survival, or at reducing their chance of getting a disease in the first place. This article proposes a smoothed version of the "Lassosum" penalty used to fit polygenic risk scores and integrated risk models. The smoothing allows one to obtain explicit gradients everywhere for efficient minimization of the Lassosum objective function while guaranteeing bounds on the accuracy of the fit. An experimental section on both Alzheimers disease and COPD (chronic obstructive pulmonary disease) demonstrates the increased accuracy of the proposed smoothed Lassosum penalty compared to the original Lassosum algorithm, allowing it to draw equal with state-of-the-art methodology such as LDpred2 when evaluated via the AUC (area under the ROC curve) metric.

bioinformatics↗

bioRxiv2020

Mutations in SARS-CoV-2 spike protein and RNA polymerase complex are associated with COVID-19 mortality risk

BackgroundSARS-CoV-2 mortality has been extensively studied in relation to host susceptibility. How sequence variations in the SARS-CoV-2 genome affect pathogenicity is poorly understood. Whole-genome sequencing (WGS) of the virus with death in SARS-CoV-2 patients is one potential method of early identification of highly pathogenic strains to target for containment. MethodsWe analyzed 7,548 single stranded RNA-genomes of SARS-CoV-2 patients in the GISAID database (Elbe and Buckland-Merrett, 2017; Shu and McCauley, 2017) and associated variants with reported patients health status from COVID-19, i.e. deceased versus non-deceased. We probed each locus of the single stranded RNA of the SARS-CoV-2 virus for direct association with host/patient mortality using a logistic regression. ResultsIn total, evaluating 29,891 loci of the viral genome for association with patient/host mortality, two loci, at 12,053bp and 25,088bp, achieved genome-wide significance (p-values of 4.09e-09 and 4.41e-23, respectively). ConclusionsMutations at 25,088bp occur in the S2 subunit of the SARS-CoV-2 spike protein, which plays a key role in viral entry of target host cells. Additionally, mutations at 12,053bp are within the ORF1ab gene, in a region encoding for the protein nsp7, which is necessary to form the RNA polymerase complex responsible for viral replication and transcription. Both mutations altered amino acid coding sequences, potentially imposing structural changes that could enhance viral infectivity and symptom severity, and may be important to consider as targets for therapeutic development. Identification of these highly significant associations, unlikely to occur by chance, may assist with COVID-19 early containment of strains that are potentially highly pathogenic.

genomics↗

bioRxiv2020

A framework to efficiently smooth L1 penalties for linear regression

Penalized linear regression approaches that include an L1 term have become an important tool in statistical data analysis. One prominent example is the least absolute shrinkage and selection operator (Lasso), though the class of L1 penalized regression operators also includes the fused and graphical Lasso, the elastic net, etc. Although the L1 penalty makes their objective function convex, it is not differentiable everywhere, motivating the development of proximal gradient algorithms such as Fista, the current gold standard in the literature. In this work, we take a different approach based on smoothing in a fixed parameter setting (the problem size n and number of parameters p are fixed). The methodological contribution of our article is threefold: (1) We introduce a unified framework to compute closed-form smooth surrogates of a whole class of L1 penalized regression problems using Nesterov smoothing. The surrogates preserve the convexity of the original (unsmoothed) objective functions, are uniformly close to them, and have closed-form derivatives everywhere for efficient minimization via gradient descent; (2) We prove that the estimates obtained with the smooth surrogates can be made arbitrarily close to the ones of the original (unsmoothed) objective functions, and provide explicitly computable a priori error bounds on the accuracy of our estimates; (3) We propose an iterative algorithm to progressively smooth the L1 penalty which increases accuracy and is virtually free of tuning parameters. The proposed methodology is applicable to a large class of L1 penalized regression operators, including all the operators mentioned above. Although the resulting estimates are typically dense, sparseness can be enforced again via thresholding. Using simulation studies, we compare our framework to current gold standards such as Fista, glmnet, gLasso, etc. Our results suggest that our proposed smoothing framework provides predictions of equal or higher accuracy than the gold standards while keeping the aforementioned theoretical guarantees and having roughly the same asymptotic runtime scaling.

bioinformatics↗

bioRxiv2020

A fast and efficient smoothing approach to LASSO regression and an application in statistical genetics: polygenic risk scores for Chronic obstructive pulmonary disease (COPD)

High dimensional linear regression problems are often fitted using Lasso approaches. Although the Lasso objective function is convex, it is not differentiable everywhere, making the use of gradient descent methods for minimization not straightforward. To avoid this technical issue, we apply Nesterov smoothing to the original (unsmoothed) Lasso objective function. We introduce a closed-form smoothed Lasso which preserves the convexity of the Lasso function, is uniformly close to the unsmoothed Lasso, and allows us to obtain closed-form derivatives everywhere for efficient and fast minimization via gradient descent. Our simulation studies are focused on polygenic risk scores using genetic data from a genome-wide association study (GWAS) for chronic obstructive pulmonary disease (COPD). We compare accuracy and run-time of our approach to the current gold standard in the literature, the FISTA algorithm. Our results suggest that the proposed methodology provides estimates with equal or higher accuracy than the FISTA algorithm while having the same asymptotic runtime scaling. The proposed methodology is implemented in the R-package smoothedLasso, available on the Comprehensive R Archive Network (CRAN).

bioinformatics↗

bioRxiv2020

locStra: Fast analysis of regional/global stratification in whole genome sequencing (WGS) studies

locStra is an R-package for the analysis of regional and global population stratification in whole genome sequencing studies, where regional stratification refers to the substructure defined by the loci in a particular region on the genome. Population substructure can be assessed based on the genetic covariance matrix, the genomic relationship matrix, and the unweighted/weighted genetic Jaccard similarity matrix. Using a sliding window approach, the regional similarity matrices are compared to the global ones, based on user-defined window sizes and metrics, e.g. the correlation between regional and global eigenvectors. An algorithm for the specification of the window size is provided. As the implementation fully exploits sparse matrix algebra and is written in C++, the analysis is highly efficient. Even on single cores, for realistic study sizes (several thousand subjects, several million RVs per subject), the runtime for the genome-wide computation of all regional similarity matrices does typically not exceed one hour, enabling an unprecedented investigation of regional stratification across the entire genome. The package is applied to three WGS studies, illustrating the varying patterns of regional substructure across the genome and its beneficial effects on association testing.

genetics↗

bioRxiv2017

Proper Conditional Analysis in the Presence of Missing Data Identified Novel Independently Associated Low Frequency Variants in Nicotine Dependence Genes

Meta-analysis of genetic association studies increases sample size and the power for mapping complex traits. Existing methods are mostly developed for datasets without missing values. In practice, genotype imputation is not always effective, e.g. when targeted genotyping/sequencing assays are used or when the un-typed genetic variant is rare. Therefore, contributed summary statistics often contain missing values. Naive extensions of existing methods either replace missing summary statistics with 0 or discard studies with missing data. These approaches can bias genetic effect estimates and lead to seriously inflated type-I or II errors in conditional analysis, which is a critical tool for identifying independently associated variants.\n\nTo address this challenge and complement imputation methods, we developed a method to combine summary statistics across participating studies and consistently estimate joint effects, even when the contributed summary statistics contain large amount of missing values. Based on this estimator, we propose a score statistic we call PCBS (partial correlation based score statistic) for conditional analysis of single-variant and gene-level associations. Through extensive analysis of simulated and real data, we showed that the new method produces well-calibrated type-I errors and is substantially more powerful than existing approaches. We applied the proposed approach to analyze the CHRNA5-CHRNB4-CHRNA3 locus in a large-scale meta-analysis for cigarettes-per-day. Using the new method, we identified three novel variants, independent of known association signals, which were otherwise missed by alternative methods. Together, the phenotypic variance explained by these variants is .46%, improving that of previously reported associations by 17%. These findings illustrate the extent of locus allelic heterogeneity and can help pinpoint causal variants.\n\nAUTHOR SUMMARYIt is of great interest to estimate the joint and conditional effects of multiple correlated variants from large scale meta-analysis, in order to fine map causal variants and understand the genetic architecture for complex traits. The contributed summary statistics from participating studies in a meta-analysis often contain missing values, as the imputation methods are not often effective, especially when the underlying genetic variant is rare or the participating studies use targeted genotyping array that is not suitable for imputation. Existing meta-analysis methods do not properly handle missing data, and can incorrectly estimate correlations between score statistics. As a result, they can produce highly biased estimates of joint effects and highly inflated type-I errors for conditional analysis, which will in turn result in overestimated phenotypic variance explained and incorrect identification of causal variants. We systematically evaluated this bias and proposed a novel partial correlation based score statistic. The new statistic has valid type-I errors for conditional analysis and much higher power than the existing methods, even when the contributed summary statistics in the meta-analysis contain a large fraction of missing values. We expect this method to be highly useful in the sequencing age for complex trait genetics.

genetics↗

bioRxiv2017

Association Analysis and Meta-Analysis of Multi-allelic Variants for Large Scale Sequence Data

MotivationThere is great interest to understand the impact of rare variants in human diseases using large sequence datasets. In deep sequences datasets of >10,000 samples, [~]10% of the variant sites are observed to be multi-allelic. Many of the multi-allelic variants have been shown to be functional and disease relevant. Proper analysis of multi-allelic variants is critical to the success of a sequencing study, but existing methods do not properly handle multi-allelic variants and can produce highly misleading association results.\n\nResultsWe propose novel methods to encode multi-allelic sites, conduct single variant and gene-level association analyses, and perform meta-analysis for multi-allelic variants. We evaluated these methods through extensive simulations and the study of a large meta-analysis of [~]18,000 samples on the cigarettes-per-day phenotype. We showed that our joint modeling approach provided an unbiased estimate of genetic effects, greatly improved the power of single variant association tests, and enhanced gene-level tests over existing approaches.\n\nAvailabilitySoftware packages implementing these methods are available at (https://github.com/zhanxw/rvtests http://genome.sph.umich.edu/wiki/RareMETAL).\n\nContactxiaowei.zhan@utsouthwestem.edu; dajiang.liu@psu.edu

bioinformatics↗