pycondiments 0.1.0

Pure-Python port of condiments — differential trajectory analysis across conditions (Roux de Bezieux et al. Nat Commun 2024).

py-condiments

A Python port of condiments (Roux de Bezieux et al., Nature Communications 2024) — differential-trajectory analysis between conditions in single-cell RNA-seq.

• AnnData-compatible
• 7/7 R exports ported
• imbalance_score Pearson = 1.0000 vs R on canonical toy fixture

Install

pip install pycondiments

Quick-start

import pycondiments as cd

# Given a Slingshot pseudotime + cellWeights + condition labels per cell:
imb = cd.imbalance_score(rd, conditions, k=10, smooth=10)
ptest = cd.progressionTest(pseudotime, conditions, cellWeights=cellWeights)
ttest = cd.topologyTest(pseudotime, cellWeights, conditions)

Function map

Python	R	Status
imbalance_score	imbalance_score	✅ Pearson 1.000 vs R
progressionTest	progressionTest	✅ both agree on significance
differentiationTest	fateSelectionTest	✅ (multi-lineage only)
topologyTest	topologyTest	🟡 simplified (v0.1)
weights_from_pst	weights_from_pst	✅
merge_sds	merge_sds	✅
create_differential_topology	create_differential_topology	✅ test-data helper

Known limitations (v0.1)

1. topologyTest is approximate: uses χ² on dominant-lineage contingency rather than re-fitting Slingshot per condition. Strong topology changes will be missed.
2. No GAM smoothing in imbalance_score: R uses mgcv s(); we use kNN-average. Both give similar smoothed z-scores.
3. No multi-condition support: v0.1 supports 2-condition tests only. Multi-condition extension is on the v0.2 roadmap.

Citation

Roux de Bezieux, H. et al. Trajectory inference across multiple conditions with condiments. Nature Communications 15, 1281 (2024).

License

MIT.