omicverse-single-cell-differential-expression

What problem does it solve?

Turn interactive notebook DEG workflows into a repeatable, triggerable analysis that compares conditions inside one or more cell types in an AnnData object, removing guesswork about backend choice, matrix provenance, and result validation.

Core Features & Use Cases

• Subset AnnData by cell-type and condition, then run differential expression using Wilcoxon, t-test, or memento-de backends.
• Auto-detect and prefer raw counts from adata.raw or a counts layer, attempt count recovery when appropriate, and downsample large subsets via max_cells.
• Produce standardized DEG tables augmented with log2FC, pvalue, padj, baseMean, pct_ctrl, pct_test, and pct_diff for downstream plotting and reporting.
• Use Case: compare Control vs Salmonella within a specific epithelial subpopulation, verify result columns, and optionally rerun with memento-de for count-aware inference.

Quick Start

Run a per-cell-type differential expression comparing Control and Salmonella inside an AnnData object using the wilcoxon backend and default validation checks.