omicverse-single-cell-differential-expression

Single-cell differential expression compares condition-versus-control groups within selected cell types using AnnData objects. This workflow subsets specific cell populations, validates control and test labels, and applies statistical backends like Wilcoxon, t-test, or memento-de to generate standardized DEG tables with log2FC and adjusted p-values.

Memento-de is a statistical backend for single-cell differential expression analysis that processes raw counts from AnnData objects. It works alongside Wilcoxon and t-test methods to calculate expression changes and significance metrics between condition groups within specific cell types.

OmicVerse DEG workflows support optional raw-count detection and max_cells downsampling control for large AnnData datasets. These features manage computational load during condition-versus-condition comparisons while ensuring post-run validation of expression and significance columns.

Comparing Wilcoxon, t-test, and memento-de backends involves evaluating their statistical assumptions for single-cell differential expression. This workflow standardizes the comparison by outputting uniform DEG tables with log2FC, p-values, and expression-percentage metrics regardless of the chosen backend.

Single-cell DEG analysis requires explicit control and test labels to perform valid condition-versus-condition comparisons. Without proper group validation, the statistical backends cannot accurately calculate log2FC, p-values, and adjusted significance values for downstream interpretation.

An AnnData object is required to run reusable single-cell differential expression workflows. The pipeline depends on AnnData structures for cell-type subsetting, raw-count handling, and generating standardized DEG tables with expression-percentage metrics.