arXiv subjects

Aaron J. Molstad

Publications and source records attributed to Aaron J. Molstad.

At least 19 recordsLinked to original sources

arXiv2025

A unified approach to spatial domain detection and cell-type deconvolution in spot-based spatial transcriptomics

Popular technologies for generating spatially resolved transcriptomic data measure gene expression at the resolution of a "spot", i.e., a small tissue region 55 microns in diameter. Each spot can contain many cells of different types. In typical analyses, researchers are interested in using these data to identify and profile discrete spatial domains in the tissue. In this paper, we propose a new method, DUET, that simultaneously identifies discrete spatial domains and estimates each spot's cell-type proportion. This allows the identified spatial domains to be characterized in terms of the cell type proportions, which affords interpretability and biological insight. DUET utilizes a constrained version of model-based convex clustering, and as such, can accommodate Poisson, negative binomial, normal, and other types of expression data. Through simulation studies and multiple applications, we show that DUET can achieve better clustering and deconvolution performance than existing methods.

stat.AP↗

arXiv2025

A direct approach to tree-guided feature aggregation for high-dimensional regression

In high-dimensional linear models, sparsity is often exploited to reduce variability and achieve parsimony. Equi-sparsity, where one assumes that predictors can be aggregated into groups sharing the same effects, is an alternative parsimonious structure that can be more suitable in certain applications. Previous work has clearly demonstrated the benefits of exploiting equi-sparsity in the presence of ``rare features'' (Yan and Bien 2021). In this work, we propose a new tree-guided regularization scheme for simultaneous estimation and feature aggregation. Unlike existing methods, our estimator avoids synthetic overparameterization and its detrimental effects. Even though our penalty is applied to hierarchically overlapped groups, we show that its proximal operator can be solved with a one-pass, non-iterative algorithm. Novel techniques are developed to study the finite-sample error bound of this seminorm-induced regularizer under least squares and binomial deviance losses. Theoretically, compared to existing methods, the proposed method offers a faster or equivalent rate depending on the true equi-sparisty structure. Extensive simulation studies verify these findings. Finally, we illustrate the usefulness of the proposed method with an application to a microbiome dataset, where we conduct post-selection inference on the aggregated features' effects.

stat.ME↗

arXiv2025

The Why and How of Convex Clustering

This survey reviews a clustering method based on solving a convex optimization problem. Despite the plethora of existing clustering methods, convex clustering has several uncommon features that distinguish it from prior art. The optimization problem is free of spurious local minima, and its unique global minimizer is stable with respect to all its inputs, including the data, a tuning parameter, and weight hyperparameters. Its single tuning parameter controls the number of clusters and can be chosen using standard techniques from penalized regression. We give intuition into the behavior and theory for convex clustering as well as practical guidance. We highlight important algorithms and discuss how their computational costs scale with the problem size. Finally, we highlight the breadth of its uses and flexibility to be combined and integrated with other inferential methods.

stat.ME↗

arXiv2025

Universal inference for variance components

We consider universal inference in variance components models, focusing on settings where the parameter is near or at the boundary of the parameter set. Two cases, which are not handled by existing state-of-the-art methods, are of particular interest: (i) inference on a variance component when other variance components are near or at the boundary, and (ii) inference on near-unity proportions of variability, that is, one variance component divided by the sum of all variance components. Case (i) is relevant, for example, for the construction of componentwise confidence intervals, as often used by practitioners. Case (ii) is particularly relevant when making inferences about heritability in modern genetics. For both cases, we show how to construct confidence intervals that are uniformly valid in finite samples. We propose algorithms which, by exploiting the structure of variance components models, lead to substantially faster computing than naive implementations of universal inference. The usefulness of the proposed methods is illustrated by simulations and a data example with crossed random effects, which are known to be complicated for conventional inference procedures.

stat.ME↗

arXiv2024

Fast and reliable confidence intervals for a variance component

We show that confidence intervals in a variance component model, with asymptotically correct uniform coverage probability, can be obtained by inverting certain test-statistics based on the score for the restricted likelihood. The results apply in settings where the variance is near or at the boundary of the parameter set. Simulations indicate the proposed test-statistics are approximately pivotal and lead to confidence intervals with near-nominal coverage even in small samples. We illustrate our methods' application in spatially-resolved transcriptomics where we compute approximately 15,000 confidence intervals, used for gene ranking, in less than 4 minutes. In the settings we consider, the proposed method is between two and 28,000 times faster than popular alternatives, depending on how many confidence intervals are computed.

stat.ME↗

arXiv2024

Subspace decompositions for association structure learning in multivariate categorical response regression

Modeling the complex relationships between multiple categorical response variables as a function of predictors is a fundamental task in the analysis of categorical data. However, existing methods can be difficult to interpret and may lack flexibility. To address these challenges, we introduce a penalized likelihood method for multivariate categorical response regression that relies on a novel subspace decomposition to parameterize interpretable association structures. Our approach models the relationships between categorical responses by identifying mutual, joint, and conditionally independent associations, which yields a linear problem within a tensor product space. We establish theoretical guarantees for our estimator, including error bounds in high-dimensional settings, and demonstrate the method's interpretability and prediction accuracy through comprehensive simulation studies.

stat.ME↗

arXiv2024

Smooth and shape-constrained quantile distributed lag models

Exposure to environmental pollutants during the gestational period can significantly impact infant health outcomes, such as birth weight and neurological development. Identifying critical windows of susceptibility, which are specific periods during pregnancy when exposure has the most profound effects, is essential for developing targeted interventions. Distributed lag models (DLMs) are widely used in environmental epidemiology to analyze the temporal patterns of exposure and their impact on health outcomes. However, traditional DLMs focus on modeling the conditional mean, which may fail to capture heterogeneity in the relationship between predictors and the outcome. Moreover, when modeling the distribution of health outcomes like gestational birthweight, it is the extreme quantiles that are of most clinical relevance. We introduce two new quantile distributed lag model (QDLM) estimators designed to address the limitations of existing methods by leveraging smoothness and shape constraints, such as unimodality and concavity, to enhance interpretability and efficiency. We apply our QDLM estimators to the Colorado birth cohort data, demonstrating their effectiveness in identifying critical windows of susceptibility and informing public health interventions.

stat.AP↗

arXiv2024

Conditional probability tensor decompositions for multivariate categorical response regression

In many modern regression applications, the response consists of multiple categorical random variables whose probability mass is a function of a common set of predictors. In this article, we propose a new method for modeling such a probability mass function in settings where the number of response variables, the number of categories per response, and the dimension of the predictor are large. Our method relies on a functional probability tensor decomposition: a decomposition of a tensor-valued function such that its range is a restricted set of low-rank probability tensors. This decomposition is motivated by the connection between the conditional independence of responses, or lack thereof, and their probability tensor rank. We show that the model implied by such a low-rank functional probability tensor decomposition can be interpreted in terms of a mixture of regressions and can thus be fit using maximum likelihood. We derive an efficient and scalable penalized expectation maximization algorithm to fit this model and examine its statistical properties. We demonstrate the encouraging performance of our method through both simulation studies and an application to modeling the functional classes of genes.

stat.ME↗

arXiv2024

Direct covariance matrix estimation with compositional data

Compositional data arise in many areas of research in the natural and biomedical sciences. One prominent example is in the study of the human gut microbiome, where one can measure the relative abundance of many distinct microorganisms in a subject's gut. Often, practitioners are interested in learning how the dependencies between microbes vary across distinct populations or experimental conditions. In statistical terms, the goal is to estimate a covariance matrix for the (latent) log-abundances of the microbes in each of the populations. However, the compositional nature of the data prevents the use of standard estimators for these covariance matrices. In this article, we propose an estimator of multiple covariance matrices which allows for information sharing across distinct populations of samples. Compared to some existing estimators, which estimate the covariance matrices of interest indirectly, our estimator is direct, ensures positive definiteness, and is the solution to a convex optimization problem. We compute our estimator using a proximal-proximal gradient descent algorithm. Asymptotic properties of our estimator reveal that it can perform well in high-dimensional settings. Through simulation studies, we demonstrate that our estimator can outperform existing estimators. We show that our method provides more reliable estimates than competitors in an analysis of microbiome data from subjects with chronic fatigue syndrome.

stat.ME↗

arXiv2024

Heterogeneity-aware integrative regression for ancestry-specific association studies

Ancestry-specific proteome-wide association studies (PWAS) based on genetically predicted protein expression can reveal complex disease etiology specific to certain ancestral groups. These studies require ancestry-specific models for protein expression as a function of SNP genotypes. In order to improve protein expression prediction in ancestral populations historically underrepresented in genomic studies, we propose a new penalized maximum likelihood estimator for fitting ancestry-specific joint protein quantitative trait loci models. Our estimator borrows information across ancestral groups, while simultaneously allowing for heterogeneous error variances and regression coefficients. We propose an alternative parameterization of our model which makes the objective function convex and the penalty scale invariant. To improve computational efficiency, we propose an approximate version of our method and study its theoretical properties. Our method provides a substantial improvement in protein expression prediction accuracy in individuals of African ancestry, and in a downstream PWAS analysis, leads to the discovery of multiple associations between protein expression and blood lipid traits in the African ancestry population.

stat.AP↗

arXiv2023

Multiresolution categorical regression for interpretable cell type annotation

In many categorical response regression applications, the response categories admit a multiresolution structure. That is, subsets of the response categories may naturally be combined into coarser response categories. In such applications, practitioners are often interested in estimating the resolution at which a predictor affects the response category probabilities. In this article, we propose a method for fitting the multinomial logistic regression model in high dimensions that addresses this problem in a unified and data-driven way. In particular, our method allows practitioners to identify which predictors distinguish between coarse categories but not fine categories, which predictors distinguish between fine categories, and which predictors are irrelevant. For model fitting, we propose a scalable algorithm that can be applied when the coarse categories are defined by either overlapping or nonoverlapping sets of fine categories. Statistical properties of our method reveal that it can take advantage of this multiresolution structure in a way existing estimators cannot. We use our method to model cell type probabilities as a function of a cell's gene expression profile (i.e., cell type annotation). Our fitted model provides novel biological insights which may be useful for future automated and manual cell type annotation methodology.

stat.ME↗

arXiv2023

Dimension reduction for integrative survival analysis

We propose a constrained maximum partial likelihood estimator for dimension reduction in integrative (e.g., pan-cancer) survival analysis with high-dimensional covariates. We assume that for each population in the study, the hazard function follows a distinct Cox proportional hazards model. To borrow information across populations, we assume that all of the hazard functions depend only on a small number of linear combinations of the predictors. We estimate these linear combinations using an algorithm based on "distance-to-set" penalties. This allows us to impose both low-rankness and sparsity. We derive asymptotic results which reveal that our regression coefficient estimator is more efficient than fitting a separate proportional hazards model for each population. Numerical experiments suggest that our method outperforms related competitors under various data generating models. We use our method to perform a pan-cancer survival analysis relating protein expression to survival across 18 distinct cancer types. Our approach identifies six linear combinations, depending on only 20 proteins, which explain survival across the cancer types. Finally, we validate our fitted model on four external datasets and show that our estimated coefficients can lead to better prediction than popular competitors.

stat.ME↗

arXiv2022

A Convex-Nonconvex Strategy for Grouped Variable Selection

This paper deals with the grouped variable selection problem. A widely used strategy is to augment the negative log-likelihood function with a sparsity-promoting penalty. Existing methods include the group Lasso, group SCAD, and group MCP. The group Lasso solves a convex optimization problem but is plagued by underestimation bias. The group SCAD and group MCP avoid this estimation bias but require solving a nonconvex optimization problem that may be plagued by suboptimal local optima. In this work, we propose an alternative method based on the generalized minimax concave (GMC) penalty, which is a folded concave penalty that maintains the convexity of the objective function. We develop a new method for grouped variable selection in linear regression, the group GMC, that generalizes the strategy of the original GMC estimator. We present an efficient algorithm for computing the group GMC estimator and also prove properties of the solution path to guide its numerical computation and tuning parameter selection in practice. We establish error bounds for both the group GMC and original GMC estimators. A rich set of simulation studies and a real data application indicate that the proposed group GMC approach outperforms existing methods in several different aspects under a wide array of scenarios.

stat.ME↗

arXiv2022

New insights for the multivariate square-root lasso

We study the multivariate square-root lasso, a method for fitting the multivariate response linear regression model with dependent errors. This estimator minimizes the nuclear norm of the residual matrix plus a convex penalty. Unlike existing methods that require explicit estimates of the error precision (inverse covariance) matrix, the multivariate square-root lasso implicitly accounts for error dependence and is the solution to a convex optimization problem. We establish error bounds which reveal that like the univariate square-root lasso, the multivariate square-root lasso is pivotal with respect to the unknown error covariance matrix. In addition, we propose a variation of the alternating direction method of multipliers algorithm to compute the estimator and discuss an accelerated first order algorithm that can be applied in certain cases. In both simulation studies and a genomic data application, we show that the multivariate square-root lasso can outperform more computationally intensive methods that require explicit estimation of the error precision matrix.

stat.ME↗

arXiv2022

Mixed-type multivariate response regression with covariance estimation

We propose a new method for multivariate response regression and covariance estimation when elements of the response vector are of mixed types, for example some continuous and some discrete. Our method is based on a model which assumes the observable mixed-type response vector is connected to a latent multivariate normal response linear regression through a link function. We explore the properties of this model and show its parameters are identifiable under reasonable conditions. We impose no parametric restrictions on the covariance of the latent normal other than positive definiteness, thereby avoiding assumptions about unobservable variables which can be difficult to verify in practice. To accommodate this generality, we propose a novel algorithm for approximate maximum likelihood estimation that works "off-the-shelf" with many different combinations of response types, and which scales well in the dimension of the response vector. Our method typically gives better predictions and parameter estimates than fitting separate models for the different response types and allows for approximate likelihood ratio testing of relevant hypotheses such as independence of responses. The usefulness of the proposed method is illustrated in simulations; and one biomedical and one genomic data example.

stat.ME↗

arXiv2022

A likelihood-based approach for multivariate categorical response regression in high dimensions

We propose a penalized likelihood method to fit the bivariate categorical response regression model. Our method allows practitioners to estimate which predictors are irrelevant, which predictors only affect the marginal distributions of the bivariate response, and which predictors affect both the marginal distributions and log odds ratios. To compute our estimator, we propose an efficient first order algorithm which we extend to settings where some subjects have only one response variable measured, i.e., the semi-supervised setting. We derive an asymptotic error bound which illustrates the performance of our estimator in high-dimensional settings. Generalizations to the multivariate categorical response regression model are proposed. Finally, simulation studies and an application in pan-cancer risk prediction demonstrate the usefulness of our method in terms of interpretability and prediction accuracy. An R package implementing the proposed method is available for download at github.com/ajmolstad/BvCategorical.

stat.ME↗

arXiv2022

Scalable algorithms for semiparametric accelerated failure time models in high dimensions

Semiparametric accelerated failure time (AFT) models are a useful alternative to Cox proportional hazards models, especially when the assumption of constant hazard ratios is untenable. However, rank-based criteria for fitting AFT models are often non-differentiable, which poses a computational challenge in high-dimensional settings. In this article, we propose a new alternating direction method of multipliers algorithm for fitting semiparametric AFT models by minimizing a penalized rank-based loss function. Our algorithm scales well in both the number of subjects and number of predictors; and can easily accommodate a wide range of popular penalties. To improve the selection of tuning parameters, we propose a new criterion which avoids some common problems in cross-validation with censored responses. Through extensive simulation studies, we show that our algorithm and software is much faster than existing methods (which can only be applied to special cases), and we show that estimators which minimize a penalized rank-based criterion often outperform alternative estimators which minimize penalized weighted least squares criteria. Application to nine cancer datasets further demonstrates that rank-based estimators of semiparametric AFT models are competitive with estimators assuming proportional hazards model in high-dimensional settings, whereas weighted least squares estimators are often not. A software package implementing the algorithm, along with a set of auxiliary functions, is available for download at github.com/ajmolstad/penAFT.

stat.ME↗

arXiv2021

An explicit mean-covariance parameterization for multivariate response linear regression

We develop a new method to fit the multivariate response linear regression model that exploits a parametric link between the regression coefficient matrix and the error covariance matrix. Specifically, we assume that the correlations between entries in the multivariate error random vector are proportional to the cosines of the angles between their corresponding regression coefficient matrix columns, so as the angle between two regression coefficient matrix columns decreases, the correlation between the corresponding errors increases. We highlight two models under which this parameterization arises: the latent variable reduced-rank regression model and the errors-in-variables regression model. We propose a novel non-convex weighted residual sum of squares criterion which exploits this parameterization and admits a new class of penalized estimators. The optimization is solved with an accelerated proximal gradient descent algorithm. Our method is used to study the association between microRNA expression and cancer drug activity measured on the NCI-60 cell lines. An R package implementing our method, MCMVR, is available at github.com/ajmolstad/MCMVR.

stat.ME↗