Trajectory, spatial, and cohort
The disease_workflows pipeline combines immune trajectory interpretation, spatial niche summaries, and disease cohort comparison.
Trajectory
The pipeline maps immune states onto simple immune axes:
- naive/memory
- effector
- exhausted/terminal
- B cell maturation
- myeloid polarization
It also joins clonotypes to pseudotime for lineage tracing summaries.
Spatial
Spatial rows can include x, y, cell type, immune state, and cytokine score. Outputs include:
- immune niche summaries
- TLS-like region counts
- proximity scoring
- cytokine projection rows
- spatial pattern classification: inflamed, tolerogenic, suppressive, or mixed
- graph neighborhoods from coordinate radius edges
- observed-vs-expected neighborhood enrichment
- local cytokine and immune-state gradients
- tissue/pathology region summaries when region labels are supplied
- explicit spatial edge payloads for downstream overlays
TLS-like candidate records include B-cell, T-cell, dendritic, plasma, density, and cytokine evidence fields when available.
Cohort
Cohort rows can include disease label, disease activity score, medication, timepoint, vaccination/infection history, condition, and response. Outputs include composition proportions, logit proportions, disease presets, and cross-disease deltas.