EasyDecon Documentation
easydecon provides marker-gene based deconvolution utilities for
single-cell references and spatial transcriptomics. It reads or generates
marker tables, scores marker evidence in spatial locations, combines Phase 1
priors with Phase 2 likelihoods, and returns both probabilistic support
matrices and hard assignments.
The package is designed around explicit, inspectable matrices rather than a black-box cell-fraction model. Marker quality, gene identifier overlap, spatial resolution, and the chosen scoring method all affect interpretation, so the guides below separate the basic workflow from advanced marker roles, refinement, candidate pruning, and niche detection.
Start here
Installation and quickstart for installation and a minimal workflow.
How easydecon works for the conceptual flow from markers to assignments.
Marker inputs and table schema for supported marker sources and table schema.
Understanding EasyDeconResult for result matrices and interpretation.
Core workflow
Marker preparation
Advanced workflows
Validation and reference
- Synthetic validation
- Optional bin2cell segmentation helper
- Troubleshooting
- No marker genes remain after filtering
- Gene names do not overlap
- posterior_df is None
- All rows are unassigned
- Too many ties
- Phase 1 priors are all zero
- UCell scores are all zero
- No negative markers are present
- Reference marker generation returns no markers
- PyDESeq2 raw-count validation fails
- Insufficient pseudobulk replicates
- Missing spatial coordinates
- Candidate pruning incompatibility
- refine_group finds no eligible locations
- Optional SpatialData dependency missing
- Documentation build import failures
- High runtime for permutation Phase 1
- Memory concerns
- Glossary
- API reference