CellPainting-Skills¶

cellpainting_skills is the public task package of CellPainting-Claw.

Its purpose is to give both people and agents the same stable names for concrete Cell Painting tasks. Each skill is designed to produce a usable output that can stand on its own or feed the next task.

Public Model¶

A public skill in this project is:

one named task
one main result
one documented interface that can be used from the CLI, Python, or OpenClaw

The main public catalog below focuses only on the current recommended skill names. Advanced and legacy aliases still exist in the codebase for compatibility, but they are not the primary documentation path.

Main Catalog¶

The tables below are written from the user point of view:

main input is the artifact or setting the skill starts from
main result is the main output the skill is expected to leave behind
main tools shows which package or component does most of the work

Data Access¶

Skill key	Main input	Main result	Main tools
`data-inspect-availability`	a project config with data-access settings	an availability summary for the configured data sources	`boto3`, `quilt3`, `cpgdata`
`data-plan-download`	a project config plus a requested dataset or prefix	a saved download plan without downloading files yet	`boto3`, `quilt3`, `cpgdata`
`data-download`	a saved download plan or direct data request	a local download cache plus execution records	`boto3`, `quilt3`, `cpgdata`

Profiling¶

Skill key	Main input	Main result	Main tools
`cp-extract-measurements`	raw Cell Painting images plus profiling configuration	CellProfiler measurement tables such as `Image.csv`, `Cells.csv`, `Cytoplasm.csv`, and `Nuclei.csv`	`CellProfiler`
`cp-build-single-cell-table`	CellProfiler measurement tables	one merged single-cell table for downstream analysis	`cellpaint_pipeline.profiling_native`
`cyto-aggregate-profiles`	a single-cell table	an aggregated classical profile table	`pycytominer`
`cyto-annotate-profiles`	an aggregated profile table	a metadata-annotated profile table	`pycytominer`
`cyto-normalize-profiles`	an annotated profile table	a normalized profile table	`pycytominer`
`cyto-select-profile-features`	a normalized profile table	a feature-selected profile table	`pycytominer`
`cyto-summarize-classical-profiles`	one or more classical profile tables	a readable summary bundle with metadata and PCA outputs	`pandas`, `numpy`

Segmentation¶

Skill key	Main input	Main result	Main tools
`cp-prepare-segmentation-inputs`	image metadata plus segmentation settings	the load-data table used by segmentation	`cellpaint_pipeline.segmentation_native`
`cp-extract-segmentation-artifacts`	a segmentation load-data table plus the selected segmentation `.cppipe`	masks, labels, outlines, and segmentation measurement tables	`CellProfiler`
`cp-generate-segmentation-previews`	segmentation inputs or a segmentation workflow root	preview PNGs for quick review	`Pillow`, `numpy`
`crop-export-single-cell-crops`	a completed segmentation workflow root	masked or unmasked single-cell crop stacks plus a crop manifest	`cellpaint_pipeline.segmentation_native`

Deep Features¶

Skill key	Main input	Main result	Main tools
`dp-export-deep-feature-inputs`	segmentation outputs or crop exports	DeepProfiler-ready metadata and location files	DeepProfiler export helpers
`dp-build-deep-feature-project`	a prepared deep-feature input bundle	a runnable DeepProfiler project directory	DeepProfiler project helpers
`dp-run-deep-feature-model`	a prepared DeepProfiler project	raw DeepProfiler feature files	`DeepProfiler`
`dp-collect-deep-features`	a completed DeepProfiler project or run directory	single-cell and aggregated deep-feature tables	`pandas`, `pyarrow`
`dp-summarize-deep-features`	collected deep-feature tables	a readable summary bundle with metadata and PCA outputs	`pandas`, `numpy`