TurboCell Atlas#

TurboCell Atlas is a Python package for atlas-scale single-cell retrieval. It combines compressed candidate generation with exact reranking in the original embedding space, and it is documented as a package-style site rather than a loose collection of pages.

The documentation is written for two kinds of readers at once:

  • developers who want to understand the API, benchmark setup, and retrieval stack

  • wet-lab and collaborative researchers who need a careful explanation of what the package can do, what inputs are required, and how to read the output tables and figures

What TurboCell Atlas can do

Start with the plain-language explanation of where the package is already useful and what kind of biological questions it can answer.

What TurboCell Atlas can do
What you need before using it

Read the simplest checklist of required data, metadata, and expectations before a first run.

What you need before using TurboCell Atlas
Tutorials

Follow prose-rich walkthroughs that connect a biological question, a query, and an interpretable output.

Tutorials
API reference

Inspect the package surface, configuration objects, and search entry points.

API reference

Note

Best current success case: rare or coherent disease-state retrieval. Main current limitation: the current Python prototype still leaves TurboQuant slower than exact search in several scenarios.

Start reading here#

If you are new to the package, this is the intended reading order.

  1. What TurboCell Atlas can do

  2. What you need before using TurboCell Atlas

  3. Get started or Wet-lab guide

  4. one scenario article from Tutorials

Documentation map#

Tutorials

Guides

Reference