Gene expression and cancer classification

This paper evaluates combining expression-based surrogate variables and genotype-based principal components in differential expression analysis, finding the combined approach improves replicability and biological recall in ALS datasets.

This paper introduces gemGAT, a graph attention network for gene expression imputation across tissues, showing improved performance over regression-based methods on GTEx and ADNI data.

This paper introduces diffusion models for generating synthetic gene expression data and shows they outperform existing methods in predictive performance.

SurvMarker identifies survival-associated molecular features using PCA-based weighted scoring and provides an R package with documentation and examples.

The paper develops a method called Newmanization to reduce tissue signals in transcriptomic data and applies it to digestive tract cancer samples, showing improved clustering and separation compared to mutation data.

DeConveil extends differential gene expression analysis to account for genome aneuploidy by integrating CNV data, improving gene classification and biomarker discovery in cancer.

The paper introduces cpam, a time series analysis method for omics data that improves detection of temporal changes and changepoints, outperforming existing methods in false discovery rate control and power.

The paper develops a transfer learning model using weighted 1-norm non-parallel proximal support vector machines to identify biomarkers of bacterial infection and dissemination in mice, achieving 97% balanced accuracy.

The paper determines sample size requirements for ML classification in bulk RNA-seq data and finds that ML methods generally need larger samples than differential analysis power analysis.

This paper identifies seven distinct sets of prognostic biomarkers for predicting high-risk SKCM patients using transcriptomics data and machine learning models.

DESeq2-MultiBatch corrects batch effects in multifactorial RNA-seq experiments by leveraging DESeq2's internal model, preserving biological differences and improving data analysis.

DoReMiTra provides a unified radiation transcriptomics dataset collection with standardized metadata and pre-processed data for benchmarking and integrative analysis.

This study extends a gene pair classification method to ordinal outcomes, showing comparable performance with higher interpretability in glioblastoma and ovarian cancer.

The paper introduces MEMORY, a method to identify GEARs associated with cancer prognosis, finding mitosis-related genes in LUAD and immunity-related genes in BRCA.

CanID is a machine learning classifier for pediatric cancer subtypes using RNAseq data, achieving 99% accuracy for solid tumors and 92-93% for hematologic malignancies.

The paper updates the PRECOG resource by adding pediatric and immunotherapy datasets, enabling analysis of gene expression associations with clinical outcomes using statistical models and deconvolution.

The paper identifies shared driver genes across breast tissue transitions from normal to tumor states and TANTs to tumor subtypes, revealing their involvement in cancer-related pathways.

This paper recalibrates differential gene expression by genetic dosage variance to prioritize functionally relevant genes, showing enrichment in metabolic and regulatory pathways.

The paper introduces mehenDi, a tree-based differential testing method for RNA-Seq that accounts for inferential uncertainty, improving detection of differentially expressed transcripts and inner nodes.

This study uses unsupervised machine learning to identify temporal gene expression components in HeLa cells after cell cycle release, revealing oscillatory and transient patterns linked to cell cycle phases and cancer-related processes.

THRESHOLD evaluates gene saturation in disease progression using TCGA datasets, revealing co-regulation patterns and identifying biomarkers in bladder, prostate, and breast cancers.

The paper introduces methods to generate correlated omics data for benchmarking, showing improved performance in DESeq2 and CYCLOPS when dependencies are included.

The paper introduces fastBMC, an algorithm that accelerates the identification of bivariate monotonic classifiers, improving classification performance on clinical datasets.

This paper introduces QRscore, a non-parametric test for detecting distributional shifts in gene expression, showing it outperforms existing methods in detecting variance and mean shifts.

The paper introduces EGNF, a graph-based framework for biomarker discovery that outperforms traditional methods in classification accuracy and interpretability across cancer datasets.

This paper introduces an autoencoder with orthogonal constraints for multi-omics data integration, showing improved reconstruction and classification performance on cancer data.

The paper introduces anticlustering to balance batches and minimize batch effects in high-throughput sequencing, showing improved balance compared to existing methods.

This study uses transcriptomics and machine learning to identify signature genes and pathways in colon cancer, finding high classification accuracy and potential biomarkers.

The paper introduces AUC hyper-stability scores to improve the selection of Random Forest implementations for stable short biomarker signature discovery.