Spatial gene expression from H&E¶
An H&E slide is cheap, routine and already in the archive. A spatial transcriptomics assay is none of those things. The models documented here read the stained image and predict what a spatial assay would have measured, spot by spot, so that a question about where a gene is expressed can be asked of a slide you already have.
This site documents two ways to use them.
The Python package, on your machine. Prepare tiles from a slide, run quality control, write and read the standard result file. Nothing leaves your computer, and nothing here needs an account. Start with the Quickstart.
The hosted API. Submit a prepared sample, get a job, download an .h5ad
when it finishes. The submission model explains what a
submission is; the API reference is generated from the
service's own schema.
The hosted API is not open yet
The endpoints on this site are documented ahead of launch and the API host
does not answer. Keys cannot be minted and no runtime is accepting work.
The Python package's local path, which the Quickstart uses, works today and
needs nothing but a pip install.
Platform status: https://auroraomics.org/api/health. Which runtimes are accepting work is reported by GET / on the API itself — its lanes list is empty while none is open, and the API host does not answer at all before launch.
Where things are¶
| What | Where |
|---|---|
| This site | https://docs.auroraomics.org |
| API | https://api.auroraomics.org |
| Website | https://auroraomics.org |
| Whole site in one fetch | https://docs.auroraomics.org/llms.txt, https://docs.auroraomics.org/llms-full.txt |
llms.txt is a map of this site and llms-full.txt is the whole of it in one
file, so a reader that is a program can take it in a single request instead of
crawling.
What to read next¶
| If you want to | Read |
|---|---|
| a result on your own machine, now | Quickstart |
| to understand what you submit | A submission |
| to know what a model needs | input_spec |
| to know which genes come back | Genes |
| the caps before you hit them | Limits |
| to handle a failure properly | Errors |
| to know what is stored, and for how long | Privacy and retention |
What this is not¶
These models are research instruments. They are not a medical device, their output is a prediction and not a measurement, and neither may be used for any clinical purpose. Responsible use states this in the words of the Terms you accept, which are the ones that govern use.