Biomarker discovery

The biomarker pipeline supports translational workflows: feature selection from expression data, reduction to a predictive gene panel, and cross-validated classifier performance.

Inputs

• Expression matrix in AnnData (cells or samples as observations)
• Class labels in obs — default column name condition, configurable per run
• At least two classes with sufficient samples per class

Pipeline stages

Optional WGCNA

Co-expression network analysis runs on bulk-style matrices. On single-cell data WGCNA is skipped automatically because module structure is unreliable at cell-level sparsity.

The standalone WGCNA Modules step opens a dedicated module explorer with module summaries, soft-threshold diagnostics, hub genes, eigengene previews, eigengene-trait correlations, a lightweight hub-edge graph, and TSV exports.

Feature selection

Method	Description
mRMR	Minimum redundancy maximum relevance; capped input size for performance
Random forest	Importance ranking; used automatically for large gene sets
Combined	Merges rankings from multiple methods

Classification

Trains a classifier (SVM, k-NN, or random forest) with cross-validation and reports accuracy, F1, and confusion matrix summaries.

Results

Analyze → Biomarker Results shows:

• Ranked selected genes with scores
• Classifier performance metrics across CV folds, including accuracy, sensitivity, specificity, and F1
• Bounded feature-selection method consensus when multiple methods can be compared safely
• Inline disease-evidence overlap for top ranked genes
• Links from History and Interpret

Analyze → Coverage DEG Results shows:

• Genes ranked by recurrence across pairwise class comparisons
• Pairwise DE rows used to compute coverage recurrence
• Inline disease-evidence overlap for top coverage genes
• TSV exports for coverage genes, pairwise DE, and disease evidence

Analyze → WGCNA Module Explorer shows:

• Module summaries and top hubs
• Selected-module hub genes and strongest hub edges
• Eigengene-trait correlations and eigengene previews
• TSV exports for modules, hubs, eigengenes, trait correlations, and edges

KnowSeq alignment

The current implementation covers feature selection, bounded feature-method comparison, WGCNA module exploration, coverage-style DEG result review, inline disease evidence, and ML classification. Not yet implemented:

• Consistency selection across resampling
• Disease evidence retrieval from external databases
• Automatic reports connecting DEGs, biomarkers, classifier metrics, enrichment, and disease evidence

Tips

• Prefer pseudobulk or sample-level aggregation when biological replicates matter; cell-level labels inflate performance.
• Check class balance in Explore before running; severe imbalance affects CV metrics.
• Use enrichment on the selected gene list in a follow-up DE/enrichment run for pathway context.