Skip to content

API & data formats

For a scannable capability map (formats, embeddings, DR, clustering, properties, distances, and more), see Available tools.

The pages under advanced/api/ are auto-generated from Python docstrings on every docs build (docs/scripts/generate_api_docs.py). Edit docstrings in phenome/ and rebuild — there is no hand-maintained copy.

Document Description
PhenoMe pipeline Main orchestrator class for the complete workflow.
Vision model wrappers ModelWrapper (base), DinoV2ModelWrapper.
Visualization PCA, t-SNE, UMAP, centroid plots, distance distributions, correlations.
Distance computation Computing distances to reference groups.
Property computation Custom property extraction from images and masks.
Report generation ReportConfig and standalone HTML report generation.
Interactive explorer Widget-based dashboard for exploring embeddings in Jupyter.
Utilities Device helpers, transforms, image I/O, checkpoints, file discovery.
Metadata classes MetadataBase, DefaultMetadata, PathTemplateMetadata, DataFrameMetadata.
Plugins Property and metadata registries, report sections.

The Function location guide is auto-generated in the same build step from live phenome exports (docs/scripts/api_docs/function_index.py). It lists top-level imports, PhenoMe methods, and submodule exports with tier badges (public / advanced / internal) and deep links into the API pages above.

For a walkthrough of the complete pipeline, installation, and a first analysis, see the Getting started guide. For an overview of the core concepts, see Core concepts.

Document Description
HDF5 database protocol Internal checkpoint and result file structure.