The Auspice client (i.e. what you see in the web browser) requires a server behind it to
- (b) handle certain GET requests from the client, for instance "give me the zika dataset to display".
We provide a basic server to run Auspice locally -- any time you run
auspice view or
auspice develop you're running a server!
In these cases, the server is running on your computer, sending datasets and narratives, which are stored on your machine, to the Auspice client.
Alternatively, you can build your own server -- it just needs to satisfy the above two requirements.
Currently the client is able to make three requests:
|return a list of available datasets and narratives|
|return the requested dataset|
|return the requested narrative|
For instance, when you're running
auspice view if you go to localhost:4000/charon/getAvailable you'll see a list of the available datasets and narratives known to the server.
Similarly, nextstrain.org is a server which has handlers for these three API endpoints, so if you visit nextstrain.org/charon/getAvailable you'll see Nextstrain's available datasets.
See the server API for details about each of these requests.
Note that "/charon" can be changed to any address you wich by customising the client at build time. See the client-cusomisation API for more details.
The "Default" Auspice Server
The server provided with Auspice is intended to be run on your local setup. It thus scans the directories you provide when you run it in order to find datasets and narratives to serve. It has "handlers" for each of the above 3 requests -- i.e. bits of code that tell it how to handle each request.
Customising the Default Auspice Server
You can customise the default Auspice server by supplying your own handlers for each of the three GET requests.
See the API documentation for how to define these and provide them to
Writing Your Own Custom Server
The provided Auspice server also lets you define your own handlers, allowing plenty of flexibility in how requests are handled. But if you really want to implement your own server, then you only need to implement two things:
- serve the
- handle the three GET requests detailed above
Deploying via Heroku
It should be simple to deploy a custom version of auspice to any server, but we have experience using Heroku apps for this. Deploying to Heroku is straightforward, but there are a few points to note:
- You must set the config var
0.0.0.0for the app.
- You will need to either create a
npm run startscript which calls
npx auspice viewdepending on how you implement auspice)
- Make sure the datasets to be served are either (a) included in your git repo or (b) downloaded by the heroku build pipeline.
We use option (b) by specifing a npm script called