Gene expression and cancer classification

This paper introduces a personalized regression method to analyze pan-cancer gene expression data, revealing sample-specific patterns that population-level methods miss.

The paper predicts missing phenotype annotations for public RNA-seq data using well-annotated datasets, enabling better analysis of large-scale genomic data.

This paper introduces a nonparametric model, DPGP, to cluster gene expression time series data and demonstrates its effectiveness using simulated and real datasets.

The paper improves PCA for biological data by automating a graphical Bayesian method to identify significant principal components with biological interpretations, applied to proteomics data from acute myeloid leukemia patients.

The paper develops a deep learning-based risk stratification model for lung adenocarcinoma using gene coexpression networks and validates its prognostic value across multiple datasets.

The paper develops an efficient feature selection workflow and R package for high-dimensional omics data, evaluated on proteomics and gene expression datasets.

MultiDCoX identifies differentially co-expressed gene sets influenced by multiple factors in transcriptomic data, demonstrating effectiveness through simulated data and breast cancer analysis.

The paper presents a framework for reanalyzing Affymetrix GeneChip data using functional regions of interest, improving probe mapping and detection of gene expression changes.

This paper trains a variational autoencoder on TCGA RNA-seq data to model cancer gene expression and identify biologically relevant latent features.

This paper discusses machine learning approaches for RNA-Seq data classification and their implementation to improve disease diagnosis and biomarker research.

This paper introduces a novel feature selection method, NSVA, to improve RNA-seq differential expression analysis by reducing false positives and enhancing robustness.

The paper introduces the L1000 platform, a high-throughput gene expression profiling method, and reports 1.3 million perturbational profiles expanding the Connectivity Map for drug mechanism discovery and functional annotation.

This paper introduces composite measurements and molecular compressed sensing to efficiently recover gene expression profiles with fewer measurements than genes.

Clusternomics identifies clusters across heterogeneous datasets by modeling both local and global structures, showing improved performance on simulated and real-world cancer datasets.

The paper develops a preprocessing and normalization pipeline for large-scale RNA-Seq data, addressing batch and tissue effects in heterogeneous datasets.

The paper develops classifiers using miRNA and DNA methylation data to predict tumor origin with high accuracy and provides a webserver for clinical use.

This paper introduces a method to determine the optimal number of independent components for reproducible transcriptomic data analysis using ICA stability rankings.

This paper introduces Bayesian Relevance Networks to improve co-expression analysis of miRNA data by accounting for uncertainty in high-throughput sequencing, showing better reproducibility and lower false discovery rates.

The paper introduces a stage-wise testing procedure for differential expression and transcript usage that provides gene-level FDR control while maintaining transcript-level resolution.

The paper introduces an R package for deconvolution of LINCS L1000 data using outlier detection and aggregate Gaussian mixture modeling, showing improved stability and performance.

This paper introduces a normalization method for genomic data that accounts for biological differences in distributions while reducing technical variation.

Tradict predicts transcriptome-wide gene abundances and transcriptional programs using 100 marker genes, showing robustness and accuracy across 23,000 RNA-Seq datasets.

This study identifies cancer types using lowly expressed microRNAs from TCGA data, achieving 91% recall after data discretization.

This paper introduces a Poisson mixed model for differential expression analysis in RNAseq, accounting for over-dispersion and sample non-independence, showing improved power in real data.

The paper introduces PatternMarkers and GWCoGAPS to identify data-driven biomarkers from whole-genome data using NMF, demonstrating their application on GTex data.

This paper introduces a Gaussian copula-based method for RNA-Seq classification that accounts for gene dependencies, improving accuracy over independence-based models.

This paper investigates data transformations and model selection for RNA-seq co-expression analysis using Gaussian mixture models and proposes an R package for implementation.

limma couples gene-wise linear models with empirical Bayes variance moderation and weights; voom extends it to RNA-seq, and added functions test differential splicing and gene sets.

DESeq2 uses empirical Bayes priors to shrink negative binomial dispersions and log fold changes, giving reproducible effect sizes and error-controlled differential expression calls for RNA-seq.