Issue with downloading models: 503 Server Error

[jkbenotmane]

Hello @celltypist-devs,

First, I would like to thank you for your work on celltypist. The tool is incredibly valuable and helpful in our research.

I recently encountered an issue when attempting to download models using the following code:

from celltypist import models
models.download_models(force_update = True)
models.models_description()

When running the above, I get the following exception:

Exception: 🛑 Cannot fetch 'https://celltypist.cog.sanger.ac.uk/models/models.json', the error is: 503 Server Error: Service Unavailable for url: https://celltypist.cog.sanger.ac.uk/models/models.json

This suggests that the server is currently unavailable, possibly due to maintenance or unexpected downtime. I attempted this at different times to rule out temporary server issues, but the problem persists.

I thought it best to bring this to your attention, in case there might be an issue with the server or the model hosting. Could you please look into this issue and provide some guidance on how it can be resolved? I'm sure many users would benefit from a fix or a workaround.

Thank you in advance for your assistance and for your continued work on celltypist.

Best Regards,

Jasim

[prete]

Hi @jkbenotmane as mentioned in #66 and #67. The server that hosts the models is currently having problems serving files. We're aware of the issue and our infrastructure team is working hard to resolve it as quickly as possible. The issues started late last night (UK time) and [hopefully] be solved during the course of the evening.

Thank you for bearing with us!

[jkbenotmane]

Thank you very much for the prompt response and your endeavours to solve the issue.

Another critic would be, that celltypist almost always fails with :

ValueError: 🛑 Invalid expression matrix, expect log1p normalized expression to 10000 counts per cell

Making it hard to debug at times.

[ChuanXu1]

@jkbenotmane, this error has been turned into a warning since version 1.2.0.

By the way, the information indicates that you need to provide a normalized expression, such as those derived by sc.pp.normalize_total(adata, target_sum = 1e4) and sc.pp.log1p(adata) if you start with the raw count matrix containing all genes.

[jkbenotmane]

@ChuanXu1 Thank you for coming back to me, I reinstalled celltypist with pip and performed log1p normalization after scVi 1e4 normalization and it solved my issue ... Thank you for curing my stupidity!