Skip to contents

SwitchBox Classifier for WHO 2022 Tumor Grading

classifier_grade_who_2022.Rd

A pre-trained SwitchBox classifier using Top Scoring Pairs (TSPs) methodology to predict bladder tumor grades according to the WHO 2022 classification system. The classifier distinguishes between low-grade (LG) and high-grade (HG) tumors incorporating molecular and morphological features.

Usage

classifier_grade_who_2022

Format

A list with 5 components:

name

Character string; classifier identifier (e.g., "140TSPs")

TSPs

Character matrix with 2 columns ("gene1", "gene2") containing gene pair identifiers for each comparison rule. Gene identifiers are symbols (e.g., "TP53", "BRCA1")

score

Named numeric vector of rule weights/importance scores, one per gene pair. Higher values indicate more discriminative rules

tieVote

Factor indicating voting behavior for tied comparisons

labels

Character vector of length 2: class labels c("LG", "HG")

Details

This classifier was trained on bladder cancer gene expression data to predict WHO 2022 molecular grades, which integrate both molecular alterations and morphological features. The rank-based approach is robust to technical variation between datasets.

Usage:

  • Scores < 0.5 indicate low-grade tumors (LG)

  • Scores ≥ 0.5 indicate high-grade tumors (HG)

Gene identifiers: Gene symbols (HUGO nomenclature)

Number of rules: Varies (check nrow(classifier_grade_who_2022$TSPs))

See also

classifier_grade_who_1999 for WHO 1999 grading, int_predict_grade to apply the classifier