arXiv subjects

Hongyu Zhao

Publications and source records attributed to Hongyu Zhao.

At least 55 records · Page 3Linked to original sources

arXiv2023

eQTL Studies: from Bulk Tissues to Single Cells

An expression quantitative trait locus (eQTL) is a chromosomal region where genetic variants are associated with the expression levels of certain genes that can be both nearby or distant. The identifications of eQTLs for different tissues, cell types, and contexts have led to better understanding of the dynamic regulations of gene expressions and implications of functional genes and variants for complex traits and diseases. Although most eQTL studies to date have been performed on data collected from bulk tissues, recent studies have demonstrated the importance of cell-type-specific and context-dependent gene regulations in biological processes and disease mechanisms. In this review, we discuss statistical methods that have been developed to enable the detections of cell-type-specific and context-dependent eQTLs from bulk tissues, purified cell types, and single cells. We also discuss the limitations of the current methods and future research opportunities.

q-bio.GN↗

arXiv2023

Bayesian subtyping for multi-state brain functional connectome with application on adolescent brain cognition

Converging evidence indicates that the heterogeneity of cognitive profiles may arise through detectable alternations in brain functions. Particularly, brain functional connectivity, measured under resting and cognitive states, characterizes the unique neuronal interconnections across large-scale brain networks. Despite an unprecedented opportunity to uncover neurobiological subtypes through clustering or subtyping analyses on multi-state functional connectivity, few existing approaches are applicable here to accommodate the network topology and unique biological architecture of functional connectivity. To address this issue, we propose an innovative Bayesian nonparametric network-variate clustering analysis to uncover subgroups with homogeneous brain functional network patterns integrating different cognitive states. In light of the existing neuroscience literature, we assume there are unknown state-specific modular structures within functional connectivity and simultaneously impose selection to identify informative network features for defining subtypes within unsupervised learning. To further facilitate practical use, we develop a computationally efficient variational inference algorithm to perform posterior inference with satisfactory estimation accuracy. Extensive simulations show the superior clustering accuracy and plausible result of our method. Applying the method to the landmark Adolescent Brain Cognitive Development (ABCD) study, we successfully establish neurodevelopmental subtypes linked with impulsivity related behavior trait, and identify brain sub-network phenotypes under each state to signal neurobiological heterogeneity.

stat.AP↗

arXiv2022

Clustering High-dimensional Data via Feature Selection

High-dimensional clustering analysis is a challenging problem in statistics and machine learning, with broad applications such as the analysis of microarray data and RNA-seq data. In this paper, we propose a new clustering procedure called Spectral Clustering with Feature Selection (SC-FS), where we first obtain an initial estimate of labels via spectral clustering, then select a small fraction of features with the largest R-squared with these labels, i.e., the proportion of variation explained by group labels, and conduct clustering again using selected features. Under mild conditions, we prove that the proposed method identifies all informative features with high probability and achieves minimax optimal clustering error rate for the sparse Gaussian mixture model. Applications of SC-FS to four real world data sets demonstrate its usefulness in clustering high-dimensional data.

stat.ME↗

arXiv2022

Tiny-Attention Adapter: Contexts Are More Important Than the Number of Parameters

Adapter-tuning is a paradigm that transfers a pretrained language model to downstream tasks by adding and tuning a small number of new parameters. Previously proposed adapter architectures are all feed-forward neural networks. In this paper, we investigate the effectiveness of using tiny-attention -- i.e., attention with extremely small per-head dimensionality -- as adapters. Our tiny-attention adapter learns to modify the hidden states at each position directly conditioned on the hidden states at all the other positions, which is missed by the previously proposed adapters. Moreover, we view its multiple attention heads as a mixture of experts and propose to average their weights during deployment, which further reduces its inference computation cost. On the GLUE benchmark, our tiny-attention adapter outperforms the other parameter-efficient transfer learning methods as well as full fine-tuning while only updating 0.05% of the parameters. On the FewGLUE benchmark, its performance is comparable to that of GPT-3 and PET.

cs.CL↗

arXiv2022

Statistical Inference of Cell-type Proportions Estimated from Bulk Expression Data

There is a growing interest in cell-type-specific analysis from bulk samples with a mixture of different cell types. A critical first step in such analyses is the accurate estimation of cell-type proportions in a bulk sample. Although many methods have been proposed recently, quantifying the uncertainties associated with the estimated cell-type proportions has not been well studied. Lack of consideration of these uncertainties can lead to missed or false findings in downstream analyses. In this article, we introduce a flexible statistical deconvolution framework that allows a general and subject-specific covariance of bulk gene expressions. Under this framework, we propose a decorrelated constrained least squares method called DECALS that estimates cell-type proportions as well as the sampling distribution of the estimates. Simulation studies demonstrate that DECALS can accurately quantify the uncertainties in the estimated proportions whereas other methods fail. Applying DECALS to analyze bulk gene expression data of post mortem brain samples from the ROSMAP and GTEx projects, we show that taking into account the uncertainties in the estimated cell-type proportions can lead to more accurate identifications of cell-type-specific differentially expressed genes and transcripts between different subject groups, such as between Alzheimer's disease patients and controls and between males and females.

stat.ME↗

arXiv2021

A general kernel boosting framework integrating pathways for predictive modeling based on genomic data

Predictive modeling based on genomic data has gained popularity in biomedical research and clinical practice by allowing researchers and clinicians to identify biomarkers and tailor treatment decisions more efficiently. Analysis incorporating pathway information can boost discovery power and better connect new findings with biological mechanisms. In this article, we propose a general framework, Pathway-based Kernel Boosting (PKB), which incorporates clinical information and prior knowledge about pathways for prediction of binary, continuous and survival outcomes. We introduce appropriate loss functions and optimization procedures for different outcome types. Our prediction algorithm incorporates pathway knowledge by constructing kernel function spaces from the pathways and use them as base learners in the boosting procedure. Through extensive simulations and case studies in drug response and cancer survival datasets, we demonstrate that PKB can substantially outperform other competing methods, better identify biological pathways related to drug response and patient survival, and provide novel insights into cancer pathogenesis and treatment response.

stat.ML↗

arXiv2021

Variance Estimation and Confidence Intervals from High-dimensional Genome-wide Association Studies Through Misspecified Mixed Model Analysis

We study variance estimation and associated confidence intervals for parameters characterizing genetic effects from genome-wide association studies (GWAS) misspecified mixed model analysis. Previous studies have shown that, in spite of the model misspecification, certain quantities of genetic interests are estimable, and consistent estimators of these quantities can be obtained using the restricted maximum likelihood (REML) method under a misspecified linear mixed model. However, the asymptotic variance of such a REML estimator is complicated and not ready to be implemented for practical use. In this paper, we develop practical and computationally convenient methods for estimating such asymptotic variances and constructing the associated confidence intervals. Performance of the proposed methods is evaluated empirically based on Monte-Carlo simulations and real-data application.

stat.ME↗

arXiv2020

A Manifold Proximal Linear Method for Sparse Spectral Clustering with Application to Single-Cell RNA Sequencing Data Analysis

Spectral clustering is one of the fundamental unsupervised learning methods widely used in data analysis. Sparse spectral clustering (SSC) imposes sparsity to the spectral clustering and it improves the interpretability of the model. This paper considers a widely adopted model for SSC, which can be formulated as an optimization problem over the Stiefel manifold with nonsmooth and nonconvex objective. Such an optimization problem is very challenging to solve. Existing methods usually solve its convex relaxation or need to smooth its nonsmooth part using certain smoothing techniques. In this paper, we propose a manifold proximal linear method (ManPL) that solves the original SSC formulation. We also extend the algorithm to solve the multiple-kernel SSC problems, for which an alternating ManPL algorithm is proposed. Convergence and iteration complexity results of the proposed methods are established. We demonstrate the advantage of our proposed methods over existing methods via the single-cell RNA sequencing data analysis.

stat.ML↗

arXiv2020

A set of efficient methods to generate high-dimensional binary data with specified correlation structures

High dimensional correlated binary data arise in many areas, such as observed genetic variations in biomedical research. Data simulation can help researchers evaluate efficiency and explore properties of different computational and statistical methods. Also, some statistical methods, such as Monte-Carlo methods, rely on data simulation. Lunn and Davies (1998) proposed linear time complexity methods to generate correlated binary variables with three common correlation structures. However, it is infeasible to specify unequal probabilities in their methods. In this manuscript, we introduce several computationally efficient algorithms that generate high-dimensional binary data with specified correlation structures and unequal probabilities. Our algorithms have linear time complexity with respect to the dimension for three commonly studied correlation structures, namely exchangeable, decaying-product and K-dependent correlation structures. In addition, we extend our algorithms to generate binary data of general non-negative correlation matrices with quadratic time complexity. We provide an R package, CorBin, to implement our simulation methods. Compared to the existing packages for binary data generation, the time cost to generate a 100-dimensional binary vector with the common correlation structures and general correlation matrices can be reduced up to $10^5$ folds and $10^3$ folds, respectively, and the efficiency can be further improved with the increase of dimensions. The R package CorBin is available on CRAN at https://cran.r-project.org/.

stat.ME↗

arXiv2020

BoXHED: Boosted eXact Hazard Estimator with Dynamic covariates

The proliferation of medical monitoring devices makes it possible to track health vitals at high frequency, enabling the development of dynamic health risk scores that change with the underlying readings. Survival analysis, in particular hazard estimation, is well-suited to analyzing this stream of data to predict disease onset as a function of the time-varying vitals. This paper introduces the software package BoXHED (pronounced 'box-head') for nonparametrically estimating hazard functions via gradient boosting. BoXHED 1.0 is a novel tree-based implementation of the generic estimator proposed in Lee, Chen, Ishwaran (2017), which was designed for handling time-dependent covariates in a fully nonparametric manner. BoXHED is also the first publicly available software implementation for Lee, Chen, Ishwaran (2017). Applying BoXHED to cardiovascular disease onset data from the Framingham Heart Study reveals novel interaction effects among known risk factors, potentially resolving an open question in clinical literature.

stat.ML↗

arXiv2020

Inference of Dynamic Graph Changes for Functional Connectome

Dynamic functional connectivity is an effective measure for the brain's responses to continuous stimuli. We propose an inferential method to detect the dynamic changes of brain networks based on time-varying graphical models. Whereas most existing methods focus on testing the existence of change points, the dynamics in the brain network offer more signals in many neuroscience studies. We propose a novel method to conduct hypothesis testing on changes in dynamic brain networks. We introduce a bootstrap statistic to approximate the supreme of the high-dimensional empirical processes over dynamically changing edges. Our simulations show that this framework can capture the change points with changed connectivity. Finally, we apply our method to a brain imaging dataset under a natural audio-video stimulus and illustrate that we are able to detect temporal changes in brain networks. The functions of the identified regions are consistent with specific emotional annotations, which are closely associated with changes inferred by our method.

stat.AP↗

arXiv2020

A Hidden Markov Model Based Unsupervised Algorithm for Sleep/Wake Identification Using Actigraphy

Actigraphy is widely used in sleep studies but lacks a universal unsupervised algorithm for sleep/wake identification. In this study, we proposed a Hidden Markov Model (HMM) based unsupervised algorithm that can automatically and effectively infer sleep/wake states. It is an individualized data-driven approach that analyzes actigraphy from each individual respectively to learn activity characteristics and further separate sleep and wake states. We used Actiwatch and polysomnography (PSG) data from 43 individuals in the Multi-Ethnic Study of Atherosclerosis to evaluate the performance of our method. Epoch-by-epoch comparisons were made between our HMM algorithm and that embedded in the Actiwatch software (AS). The percent agreement between HMM and PSG was 85.7%, and that between AS and PSG was 84.7%. Positive predictive values for sleep epochs were 85.6% and 84.6% for HMM and AS, respectively, and 95.5% and 85.6% for wake epochs. Both methods have similar performance and tend to overestimate sleep and underestimate wake compared to PSG. Our HMM approach is able to quantify the variability in activity counts that allow us to differentiate relatively active and sedentary individuals: individuals with higher estimated variabilities tend to show more frequent sedentary behaviors. In conclusion, our unsupervised data-driven HMM algorithm achieves slightly better performance compared to the commonly used algorithm in the Actiwatch software. HMM can help expand the application of actigraphy in large-scale studies and in cases where intrusive PSG is hard to acquire or unavailable. In addition, the estimated HMM parameters can characterize individual activity patterns that can be utilized for further analysis.

stat.AP↗

arXiv2018

Phylogeny-based tumor subclone identification using a Bayesian feature allocation model

Tumor cells acquire different genetic alterations during the course of evolution in cancer patients. As a result of competition and selection, only a few subgroups of cells with distinct genotypes survive. These subgroups of cells are often referred to as subclones. In recent years, many statistical and computational methods have been developed to identify tumor subclones, leading to biologically significant discoveries and shedding light on tumor progression, metastasis, drug resistance and other processes. However, most existing methods are either not able to infer the phylogenetic structure among subclones, or not able to incorporate copy number variations (CNV). In this article, we propose SIFA (tumor Subclone Identification by Feature Allocation), a Bayesian model which takes into account both CNV and tumor phylogeny structure to infer tumor subclones. We compare the performance of SIFA with two other commonly used methods using simulation studies with varying sequencing depth, evolutionary tree size, and tree complexity. SIFA consistently yields better results in terms of Rand Index and cellularity estimation accuracy. The usefulness of SIFA is also demonstrated through its application to whole genome sequencing (WGS) samples from four patients in a breast cancer study.

stat.AP↗

arXiv2018

A pathway-based kernel boosting method for sample classification using genomic data

The analysis of cancer genomic data has long suffered "the curse of dimensionality". Sample sizes for most cancer genomic studies are a few hundreds at most while there are tens of thousands of genomic features studied. Various methods have been proposed to leverage prior biological knowledge, such as pathways, to more effectively analyze cancer genomic data. Most of the methods focus on testing marginal significance of the associations between pathways and clinical phenotypes. They can identify relevant pathways, but do not involve predictive modeling. In this article, we propose a Pathway-based Kernel Boosting (PKB) method for integrating gene pathway information for sample classification, where we use kernel functions calculated from each pathway as base learners and learn the weights through iterative optimization of the classification loss function. We apply PKB and several competing methods to three cancer studies with pathological and clinical information, including tumor grade, stage, tumor sites, and metastasis status. Our results show that PKB outperforms other methods, and identifies pathways relevant to the outcome variables.

stat.ML↗

arXiv2017

Prediction analysis for microbiome sequencing data

One primary goal of human microbiome studies is to predict host traits based on human microbiota. However, microbial community sequencing data present significant challenges to the development of statistical methods. In particular, the samples have different library sizes, the data contain many zeros and are often over-dispersed. To address these challenges, we introduce a new statistical framework, called predictive analysis in metagenomics via inverse regression (PAMIR). An inverse regression model is developed for over-dispersed microbiota counts given the trait, and then a prediction rule is constructed by taking advantage of the dimension-reduction structure in the model. An efficient Monte Carlo expectation-maximization algorithm is designed for carrying out maximum likelihood estimation. We demonstrate the advantages of PAMIR through simulations and a real data example.

stat.ME↗

arXiv2017

On the Convergence of the EM Algorithm: A Data-Adaptive Analysis

The Expectation-Maximization (EM) algorithm is an iterative method to maximize the log-likelihood function for parameter estimation. Previous works on the convergence analysis of the EM algorithm have established results on the asymptotic (population level) convergence rate of the algorithm. In this paper, we give a data-adaptive analysis of the sample level local convergence rate of the EM algorithm. In particular, we show that the local convergence rate of the EM algorithm is a random variable $\overline{K}_{n}$ derived from the data generating distribution, which adaptively yields the convergence rate of the EM algorithm on each finite sample data set from the same population distribution. We then give a non-asymptotic concentration bound of $\overline{K}_{n}$ on the population level optimal convergence rate $\overlineκ$ of the EM algorithm, which implies that $\overline{K}_{n}\to\overlineκ$ in probability as the sample size $n\to\infty$. Our theory identifies the effect of sample size on the convergence behavior of sample EM sequence, and explains a surprising phenomenon in applications of the EM algorithm, i.e. the finite sample version of the algorithm sometimes converges faster even than the population version. We apply our theory to the EM algorithm on three canonical models and obtain specific forms of the adaptive convergence theorem for each model.

math.ST↗

arXiv2017

Characterizing Spatiotemporal Transcriptome of Human Brain via Low Rank Tensor Decomposition

Spatiotemporal gene expression data of the human brain offer insights on the spa- tial and temporal patterns of gene regulation during brain development. Most existing methods for analyzing these data consider spatial and temporal profiles separately with the implicit assumption that different brain regions develop in similar trajectories, and that the spatial patterns of gene expression remain similar at different time points. Al- though these analyses may help delineate gene regulation either spatially or temporally, they are not able to characterize heterogeneity in temporal dynamics across different brain regions, or the evolution of spatial patterns of gene regulation over time. In this article, we develop a statistical method based on low rank tensor decomposition to more effectively analyze spatiotemporal gene expression data. We generalize the clas- sical principal component analysis (PCA) which is applicable only to data matrices, to tensor PCA that can simultaneously capture spatial and temporal effects. We also propose an efficient algorithm that combines tensor unfolding and power iteration to estimate the tensor principal components, and provide guarantees on their statistical performances. Numerical experiments are presented to further demonstrate the mer- its of the proposed method. An application of our method to a spatiotemporal brain expression data provides insights on gene regulation patterns in the brain.

stat.ME↗

arXiv2016

A Dirichlet Process Mixture Model for Clustering Longitudinal Gene Expression Data

Subgroup identification (clustering) is an important problem in biomedical research. Gene expression profiles are commonly utilized to define subgroups. Longitudinal gene expression profiles might provide additional information on disease progression than what is captured by baseline profiles alone. Moreover, the longitudinal gene expression data allows for intra-individual variability to be accounted for when grouping patients. Therefore, subgroup identification could be more accurate and effective with the aid of longitudinal gene expression data. However, existing statistical methods are unable to fully utilize these data for patient clustering. In this article, we introduce a novel subgroup identification method in the Bayesian setting based on longitudinal gene expression profiles. This method, called BClustLonG, adopts a linear mixed-effects framework to model the trajectory of genes over time while clustering is jointly conducted based on the regression coefficients obtained from all genes. In order to account for the correlations among genes and alleviate the high dimensionality challenges, we adopt a factor analysis model for the regression coefficients. The Dirichlet process prior distribution is utilized for the means of the regression coefficients to induce clustering. Through extensive simulation studies, we show that BClustLonG has improved performance over other clustering methods. When applied to a dataset of severely injured (burn or trauma) patients, our model is able to distinguish burn patients from trauma patients and identify interesting subgroups in trauma patients.

stat.ME↗