omicverse-single-cell-differential-abundance

Differential abundance analysis compares cell-type composition between conditions by applying scCODA Bayesian inference or milopy neighborhood testing directly on AnnData objects. You can run these workflows using OmicVerse to detect compositional shifts.

scCODA performs Bayesian inference for compositional shifts, while milopy conducts neighborhood testing on stored embeddings. This differential abundance analysis supports both statistical backends, along with legacy milo workflows, within a single-cell environment.

Yes, differential abundance analysis requires condition labels, cell-type annotations, and sample identifiers. Milo-family branches additionally require a stored embedding in your AnnData object to validate inputs before neighborhood testing.

scCODA uses Bayesian inference to analyze compositional data by modeling cell-type counts as proportions. This differential abundance analysis workflow applies scCODA to single-cell AnnData objects to identify statistically significant shifts between control and test conditions.

Yes, differential abundance analysis operates directly on single-cell AnnData objects. The workflow validates condition labels, cell-type annotations, sample identifiers, and embeddings before running scCODA or milopy backends.

milopy neighborhood testing requires a stored embedding and relies on mix_threshold and posterior sampling parameters. Differential abundance analysis with milopy validates these inputs beforehand, but branch-specific handling is necessary for accurate compositional comparison results.