Skip to content

Get started

In the web app

Sign in from your invitation, complete MFA, and check your selected organization.

  1. Data → choose a catalog and version, then select the records to analyze.
  2. Browse tools → choose a compatible module and review its inputs.
  3. Review the costed plan, then confirm execution once.
  4. Runs → open the returned Run ID to check status and results.
  5. Home → Canvas → work with your results and save the notebook. It starts private; sharing publishes only the revision you explicitly share.

Need access? Contact your organization administrator. For an error, use Report an issue in the top bar and include what you tried and the Run ID. Review the screenshot before sending.

CLI or Python

For a dev release still on TestPyPI, follow the Dev candidate installation steps in the SDK guide first. The commands below install the public PyPI release.

Use Python 3.11–3.14 in a virtual environment. Install the SDK, sign in to dev, and list the catalogs your account can access:

python -m pip install "hla-compass==5.9.2"
hla-compass auth login --env dev
hla-compass data catalogs --env dev

Complete sign-in and MFA in the browser. Use the environment from your invitation if it differs from dev. For multiple organizations, select one with hla-compass auth use-org ORGANIZATION_UUID --env dev.

For unattended scripts, create a personal API key under Profile → API keys. See authentication for storing it and making your first curl call. HLA_API_KEY takes precedence over a saved login; never put its value in code.

Read data, then run a module

Download catalog_discovery.py and run it with a provider key and catalog key from your discovery results:

python catalog_discovery.py PROVIDER_KEY CATALOG_KEY --env dev > selection.json

Use the returned version with the SDK data example. To execute an existing module, follow prepare, review, then run. Retain the exact Run ID and pinned inputs with your analysis.

Connect a chat app

Follow MCP and Chat Apps for Claude, ChatGPT or Mistral. For the optional local MCP gateway:

python -m pip install "hla-compass[mcp]==5.9.2"
hla-compass mcp gateway --help

Build a module

Developers with publication access can start a small module:

python -m pip install "hla-compass[authoring]==5.9.2"
hla-compass init peptide-summary --template no-ui
cd peptide-summary
python -m pip install -r backend/requirements.txt
hla-compass validate --strict
python -m pytest

Read the generated SKILL.md and DATA_ACCESS.md, then follow Build modules for container testing and publication.