omicverse-single-cell-differential-abundance

What problem does it solve?

Turn notebook-style differential cell-type abundance analysis into a reusable, triggerable operation so researchers can reliably compare cell-type compositions between experimental conditions without reworking exploratory code each time.

Core Features & Use Cases

• Multiple backends: Choose between scCODA posterior sampling or Milo-family neighborhood testing (milopy or milo) depending on inference needs.
• Input validation: Ensures required AnnData fields, sample identifiers, and embedding keys exist before execution to prevent common runtime errors.
• Reproducible workflows: Encapsulates constructor, run, and result collection patterns so the same analysis can be rerun, smoke-tested, and integrated into larger pipelines.
• Use Case: Compare Control versus infected samples to identify cell-type compositional shifts using either Bayesian compositional inference or neighborhood-based testing.

Quick Start

Run a differential abundance test comparing Control and Salmonella on an AnnData with sample_key batch and embedding X_pca.