Cryo2d Classify

cryo2d-classify

Scope

Reference-free 2D classification of cryo-EM particle micrographs. Includes motion correction, contrast transfer estimation, and clustering with configurable binning and angular sampling. Exposes a typed MCP toolset for batch runs against a session directory.

Expects standard cryo-EM data products (image stacks, spectra or session directories as appropriate) with pixel size / dispersion recorded in the file metadata. Out of scope: acquisition scheduling, sample preparation, and any analysis outside the segmentation stage of the workflow.

Safety

Read-only — no instrument control. The skill reads data from disk and returns results; it never moves the stage, changes dose, or alters column settings.

Usage

{
  "tool": "2d_classify",
  "arguments": {
    "input": "/data/cryo-em/session-001",
    "output": "/data/cryo-em/session-001/2d_classify"
  }
}

Returns a JSON object with ok, a result payload carrying the primary measurement or map, and diagnostics with per-item confidence. Units are SI unless stated otherwise in the result schema.

Reproducibility

Runtime: python 3.11. Pin dependencies in requirements.txt / package.json and run with a fixed seed. Reviewers can reproduce the reported numbers with:

cd skills/cryo2d-classify
pip install -r requirements.txt && python3 verify.py --fixture fixtures/sample --seed 0