Statistical Methods and Inference

Ridge Redundancy Analysis for High-Dimensional Omics Data

This paper introduces an efficient framework for ridge RDA to analyze high-dimensional omics data by addressing computational challenges through SVD and reduced memory storage.

PROLONG: Penalized Regression for Outcome guided Longitudinal Omics analysis with Network and Group constraints

PROLONG selects metabolites co-varying with tuberculosis mycobacterial load using penalized regression with network and group constraints in longitudinal omics data.

Exploring the Flexible Penalization of Bayesian Survival Analysis Using Beta Process Prior for Baseline Hazard

This paper introduces Bayesian survival analysis methods with beta process priors for variable selection in high-dimensional data, showing improved performance over existing techniques.

The winner's curse under dependence: repairing empirical Bayes using convoluted densities

The paper addresses the winners curse in statistical estimation under dependence, proposing a density convolution method to improve empirical Bayes approaches and demonstrating its effectiveness through simulations and a real-world application.

OneNet -- One network to rule them all: consensus network inference from microbiome data

OneNet infers consensus microbial interaction networks by combining seven methods, achieving higher precision and sparser networks than individual methods on synthetic and gut microbiome data.

Identifying Effect Modification of Latent Population Characteristics on Risk Factors with a Sparse Varying Coefficient Regression

The paper introduces a sparse varying coefficient regression model to capture effect modification by latent features in risk factor-disease outcome associations, showing improved performance in lung cancer stage prediction.

Enhanced Feature Selection for Microbiome Data using FLORAL: Scalable Log-ratio Lasso Regression

FLORAL improves microbial feature selection for microbiome data by addressing survival endpoints, longitudinal samples, and sequencing biases, showing better false-positive control and sensitivity in simulations and real data.

HighDimMixedModels.jl: Robust High Dimensional Mixed Models across Omics Data

This paper evaluates high dimensional mixed-effect models using simulated and real omics data, showing SCAD penalty outperforms LASSO in variable selection and estimation accuracy.

A novel model-free feature selection method with FDR control for omics-wide association analysis

This paper introduces a model-free feature selection method with FDR control for omics-wide association analysis, showing effective performance on simulated and real microbiome data.

LAVASET: Latent Variable Stochastic Ensemble of Trees. A novel ensemble method for correlated datasets.

LAVASET improves feature importance estimation in correlated datasets by incorporating correlation factors, achieving non-inferior prediction accuracy compared to Random Forests.

Supervised Rank aggregation (SRA): A novel rank aggregation approach for ensemble-based feature selection

This paper introduces a supervised rank aggregation method for ensemble-based feature selection, showing it performs as well or better than existing methods in feature selection and predictive performance.

Insights in neuronal tuning: Navigating the statistical challenges of autocorrelation and missing variables

This study addresses statistical challenges in neuronal tuning by comparing covariate selection methods using simulated and calcium data, finding that cross-validation with cyclical shift permutation test offers better power while controlling error rates.

asmbPLS: Adaptive Sparse Multi-block Partial Least Square for Survival Prediction using Multi-Omics Data

asmbPLS is a multi-omics survival prediction method that uses adaptive sparse multi-block PLS for feature selection and prediction, showing competitive performance on simulated and real data.

SpiderLearner: An ensemble approach to Gaussian graphical model estimation

SpiderLearner combines multiple Gaussian graphical model methods to create an ensemble estimate, improving performance and identifying a seven-gene risk score for ovarian cancer.

OSCAR: Optimal subset cardinality regression using the L0-pseudonorm with applications to prognostic modelling of prostate cancer

OSCAR uses an L0-pseudonorm approach to select sparse prognostic features in prostate cancer, balancing model accuracy and clinical implementation costs.

RCFGL: Rapid Condition adaptive Fused Graphical Lasso and application to modeling brain region co-expression networks

This paper introduces RCFGL, a fast method for joint estimation of condition-specific gene co-expression networks, showing improved computational efficiency and performance over existing methods.

Thresholding Gini Variable Importance with a single trained Random Forest: An Empirical Bayes Approach

RFlocalfdr thresholds variable importance from Random Forests using an empirical Bayes approach, achieving high accuracy and speed for high-dimensional datasets.

Neural Networks beyond explainability: Selective inference for sequence motifs

The paper introduces SEISM, a selective inference method to test the association between sequence motifs extracted from neural networks and phenotypes, improving analysis in regulatory genomics and GWAS.

Hi-LASSO: High-performance Python and Apache spark packages for feature selection with high-dimensional data

Hi-LASSO provides Python and Apache Spark packages for efficient feature selection in high-dimensional data, with improved performance and scalability through parallel implementation and parametric statistical tests.

Grace-AKO: A Novel and Stable Knockoff Filter for Variable Selection Incorporating Gene Network Structures

Grace-AKO improves variable selection in high-dimensional genomic data by integrating gene network structures, achieving better finite-sample FDR control and model stability.

CoxMKF: A Knockoff Filter for High-Dimensional Mediation Analysis with a Survival Outcome in Epigenetic Studies

CoxMKF identifies DNA methylation CpG sites mediating environmental exposure effects on survival outcomes while controlling FDR in high-dimensional settings.

Mixture Network Regularization of Generalized Linear Model With Application in Genomics Data

The paper proposes a weighted sparse network learning method for high-dimensional genomics data to improve prediction accuracy and model interpretability, showing robustness and good performance in simulations and a melanoma gene expression case study.

A Sparse Additive Model for High-Dimensional Interactions with an Exposure Variable

The paper introduces a method called sail to detect non-linear interactions between a key exposure variable and high-dimensional data, showing it outperforms existing methods in prediction and support recovery.

The GR2D2 Estimator for the Precision Matrices

The paper introduces the GR2D2 estimator for precision matrices, showing superior performance in high-dimensional sparse settings and identifying cancer pathways in RNA-seq data.

Model-free Prediction Test with Application to Genomics Data

The paper introduces a model-free prediction test to assess if X contributes to predicting Y, showing improved power in gene expression analyses using CITE-seq and spatial transcriptomics data.

Hybrid Rank Aggregation (HRA): A novel rank aggregation method for ensemble-based feature selection

This paper introduces a hybrid rank aggregation method for ensemble-based feature selection, showing improved robustness and performance compared to existing methods.

AIFS: A novel perspective, Artificial Intelligence infused wrapper based Feature Selection Algorithm on High Dimensional data analysis

AIFS improves feature selection in high-dimensional data by integrating AI to predict feature subset performance without building models.

BOSO: a novel feature selection algorithm for linear regression with high-dimensional data

BOSO is a novel feature selection algorithm for linear regression that outperforms existing methods in high-dimensional datasets, demonstrated through drug sensitivity prediction in cancer.

High-dimension to high-dimension screening for detecting genome-wide epigenetic regulators of gene expression

The paper introduces a screening method to identify genome-wide epigenetic regulators of gene expression by reducing dimensionality in both predictors and responses.

A conditional survival distribution-based method for censored data imputation: overcoming the hurdle in machine learning-based survival analysis

This paper introduces a method for imputing censored survival data to improve machine learning analysis in biomedical research.