Machine Learning-Derived Sex-Specific Bladder Cancer Biomarkers from RNA-Seq Data

A gene panel dataset created by Joseph R. Pizzi and last updated in April 2026. It contains machine learning-identified biomarkers for sex-specific bladder cancer development and progression, derived from gender and disease-stratified RNA-seq data. The male and female-specific panels achieved areas under the ROC curve of 0.932 and 0.914, respectively, in distinguishing cancer from non-tumor controls.

Use Cases

• Training classifiers for bladder cancer detection based on sex-specific gene expression signatures.
• Investigating sex-specific biological pathways in cancer, such as PI3K-AKT signaling or extracellular matrix reorganization.
• Validating potential biomarker genes like PRAC1, PCDH11Y, AR, PLXNA1, USP54, and PMEPA1 in external datasets.
• Building frameworks for high-performance, small gene panels (50 genes or less) for clinical biomarker discovery.

Strengths

• Specific performance metrics are provided, with male and female panels achieving AUCs of 0.932 and 0.914.
• Results are based on multiple machine learning feature selection methods (differential expression, RFE-SVM, logistic regression, random forest).
• The dataset is focused on a clinically relevant problem with identified high-impact predictor genes related to sex hormones or chromosomes.
• Data is openly shared under a CC-BY-4.0 license.

Limitations

• Row count and column-level documentation are unknown, which limits suitability assessment.
• The dataset is very small at 15.0 KB, indicating limited scope, likely containing summary results rather than raw expression data.
• Description metadata is limited; actual data structure and quality require manual inspection after download.

Provenance

Source

figshare

Collection Method

Machine learning analysis of gender and disease-stratified RNA-seq data, using multiple feature selection methods.

Freshness

Last updated 2026-04-29 05:58:40; freshness should be verified.