Gene expression and cancer classification

This study evaluates normalization procedures for transcriptome profiles of compounds, finding that batch-corrected baseline distributions work well for large datasets and offering guidance for small datasets.

This study compares the impact of different optimization methods on model performance and sparsity in transcriptomic prediction using LASSO logistic regression.

The paper demonstrates that regression coefficients and R2 from sampling distributions of the mean match those from individual data and shows applications in gene correlation analysis.

The paper provides a framework for molecular cancer subtyping that integrates quality control, normalization, batch correction, classification, and reporting, with a use case in breast cancer.

The paper introduces maGENEgerZ, an AI-based framework that identifies more expressed genes and biological insights related to breast cancer drug response using support vector machines.

The paper estimates cell-type-specific gene co-expression networks from bulk gene expression data and identifies novel co-expressions among Alzheimer's disease risk genes.

This study evaluates noninvasive biospecimens for transcriptome studies, finding hair follicles and urine cell pellets to be most promising for disease-relevant applications.

This study benchmarks RNA-seq methods across 45 labs to identify sources of variation in detecting subtle differential expression, highlighting experimental and bioinformatics factors affecting reliability.

pyComBat provides a Python implementation of ComBat and ComBat-Seq for batch effect correction in microarray and RNA-Seq data, offering similar performance with improved speed and open-source accessibility.

The paper uses a large RNA-seq dataset to determine the sample size and normalization needed for accurate phenotypic trait prediction.

BioDiscViz provides visualization tools and consensus signature selection for BioDiscML results to improve model interpretation and usability.

This study uses longitudinal RNAseq in rats to identify temporally variable genes and recovery trajectories after chemical exposure.

This paper introduces a deep generative model to perform differential gene expression analysis on single cancer samples without control samples, outperforming DESeq2 in detecting differentially expressed genes.

This paper introduces a Bayesian hierarchical mixture model with covariate adjustment to detect differentially expressed genes in paired design whole-genome data, showing improved performance over existing methods.

This study uses attention and graph transformer models to identify RNA biomarkers for NAFLD/NASH, achieving higher accuracy than baseline methods and validating top candidates in lab experiments.

The paper develops a logistic regression model with a propensity index panel of ten genes to accurately screen prostate cancer, achieving 96.36% accuracy and 0.99 AUROC.

BEST provides a web tool for comprehensive biomarker analysis in solid tumors using large-scale, multi-cohort data with standardized annotations and diverse analytical functions.

This paper introduces Local Leveling, a method to reduce bias in RNA-seq data, improving gene-gene relationship accuracy and statistical test results.

This study evaluates the replicability of RNA-Seq differential expression and enrichment analysis in cancer research, finding that underpowered experiments produce inflated effect sizes and limited replicability.

This paper evaluates methods for integrating heterogeneous genomic studies to improve phenotype prediction, finding that batch normalization with merging or integration strategies enhances classifier performance.

This study identifies potential biomarkers for Parkinson's disease using blood-based gene expression data and develops a machine learning classifier called PitArray for non-invasive detection.

This study benchmarks variational autoencoders on cancer transcriptomics data to evaluate hyperparameter effects and latent space biological relevance.

XA4C identifies Critical genes using explainable autoencoders, revealing novel biological insights in cancer gene expression data.

The paper introduces SNAIL, a normalization method that reduces false-positive co-expression associations in RNA-sequencing data, improving downstream network analysis.

clrDV detects genes with differential variability between two populations using a skew-normal distribution, showing competitive performance and faster runtime than existing methods.

TransGeneSelector uses a Transformer-based classifier to identify key genes in small plant transcriptomic datasets, outperforming traditional methods in classification and gene discovery.

The paper curates 102 breast cancer gene-expression datasets and provides a computational pipeline for data processing and metadata annotation to facilitate reuse and comparative analysis.

TranNet identifies predictors of tumor progression by analyzing gene expression differences between tumor and adjacent normal tissues using TCGA data.

The paper evaluates seven batch-effect correction algorithms for multiomic data and finds the ratio-based method is more effective, especially when batch effects are confounded with biological factors.

This study uses quantum machine learning to classify 11 cancers from 4495 gene expression samples and identifies biomarkers linked to DNA promoter methylation and patient survival.