pbcmc

pbcmc assesses uncertainty in gene expression classifiers and molecular signatures by generating synthetic subjects via permutation of gene labels to construct null distributions for subtype classifiers and quantify per-subject classification confidence.

Key Features:

• Permutation Testing: Creates synthetic datasets by permuting gene labels to assess the stability and reliability of molecular signatures using permutation testing.
• Null Distribution Construction: Builds a null distribution for each subtype classifier by testing synthetic subjects against their corresponding subtype classifier.
• Classification Confidence Measurement: Quantifies classification confidence for individual subjects based on the constructed null distributions.

Scientific Applications:

• Gene Expression Analysis: Evaluates variability and reliability of molecular signatures in gene expression classifiers for genomics research.
• Clinical Decision Support: Provides per-subject classification confidence that can inform therapy choice decisions based on gene expression data.

Methodology:

Generate synthetic subjects by permuting gene labels, test each synthetic subject against its subtype classifier, and construct null distributions per subtype to derive per-subject classification confidence.