Advices on simultaneous multi-users access to cerebroApp

[mdeloger]

Hi,

is it possible to have simultaneous multi-users access to a single cerebroApp ?
Have you already tried ?
If yes, is there a kind of "competition" when one user click somewhere and an other click elsewhere ?

We try to find a way to display one single dataset (quite huge 140k cells) to some collaborators (clinicians), without having to manage the installation of R/Rstudio/Packages to every single computer, by running the cerebroApp on an internal server.

Thank you in advance

[romanhaa]

Hi @mdeloger,

While I never optimised cerebroApp to work as efficiently as possible with multiple parallel users, the Shiny framework should keep sessions of different users separate. As far as I know, some users host cerebroApp on a server and I haven't ever heard of any conflicts relating to multiple concurrent users. Having said that, with such a large dataset, you will likely need a relatively powerful server. I would suggest to just try it and see how it your server can handle it. In principle, your idea makes a lot of sense and should work.

Best,
Roman

[davisidarta]

Hi @mdeloger

We did it in the early Cerebro days (need to update our webpage - Gosh, things change fast). We used a multi-thread docker instance running the Cerebro container with our data pre-loaded. Your idea makes sense. Note that Cerebro can get REALLY heavy when loading the full data. Consider downsampling your dataset by 10% for scalability (we shared 10k cells from a 130k dataset and it was already quite connection-wise expensive to share through web such large amounts of interactive data), or subsetting subclusters and doing re-embeddings if you have a lot of heterogeneity.

Good luck!

[mdeloger]

Thank you both for your answers :-)

As you can see here : http://macroverse.gustaveroussy.fr/ my main problem actually is that it turns in grey immediately even before loading the example dataset but no error or warnings on R side, juste :

##---------------------------------------------------------------------------##
## Launching Cerebro v1.3
##---------------------------------------------------------------------------##
Loading required package: shiny

Listening on http://0.0.0.0:80

Have you already observed this kind of behavior ?

Best

[romanhaa]

Your website works for me, even if very slowly with such a large dataset.