Repository navigation
Mantpy 1.1.0
Reusable functions for image normalization, patch encoders, graph learning and ECM analysis.
- Configure patch encoders with custom PyTorch models and export embeddings and training history.
- Fit graph classifiers with biological-unit splits and ROI-balanced loss; inspect seed-level integrated gradients.
- Summarize composition, co-occurrence, neighbourhoods, cell–ECM contacts and connected regions; reconstruct labels from probabilities.
- Reuse sparse graphs and display ID-aligned patch maps.
Validated on Python 3.11 and 3.14, with a successful package and documentation build.
Install:
pip install --upgrade mantpyFor neural analyses:
pip install --upgrade "mantpy[patch,gnn]"