search

ersilia-os / eos1pu1

GNU General Public License v3.0
0 stars 1 forks source link

New model ready for testing! #2

Closed github-actions[bot] closed 1 week ago

github-actions[bot] commented 1 month ago

This model is ready for testing. If you are assigned to this issue, please try it out using DockerHub and/or pulling from S3 and let us know if it works!

GemmaTuron commented 1 month ago

Hi @HarmonySosa and @LauraGomezjurado

@kurysauce has just incorporated this model. Can you test that it works ? This means pulling it from DockerHub and checking to pass a list of SMILES if you get results that would make sense (for example, if a probability is from 0 to 1, that you do not get all 0 or all 1)

HarmonySosa commented 1 month ago

Ok @GemmaTuron, I will work on it now - are there any other results that I should check for besides all 0 or 1?

GemmaTuron commented 1 month ago

you can always also look at the publication and get any molecule they used to test the model and see if the results are reproducible :)

HarmonySosa commented 1 month ago

Will do, thanks!

LauraGomezjurado commented 1 month ago

Perfect @GemmaTuron, will do!

HarmonySosa commented 1 month ago

Hi @GemmaTuron, I am getting “StandardModelExampleError” when I try to fetch the model, and I don’t fully understand what this means from the output. Here is the txt file with the output log when I run ersilia -v fetch eos1pu1: eos1pu1_run_error_170724.txt

Just to let you know in case it is related, earlier I had an error with DockerHub (“An unexpected error occurred” every time I tried to use DockerHub) and in the process of fixing that, I somehow lost access to Ersilia on my machine. I am not fully sure what happened, but I couldn’t access Ersilia, miniconda, or the associated packages even though I had them and was using them earlier. I reset my environment from the beginning so it is fine now, but I wanted to let you know in case my current issue is in any way related to these changes.

GemmaTuron commented 1 month ago

Hi @HarmonySosa

The model works well on my end - Make sure you have the latest Ersilia code pulled from GitHub (and that you did install it with pip install -e . ). If you cannot access conda, you have likely messed the python paths - let me know if you need help

GemmaTuron commented 1 month ago

Hi @LauraGomezjurado and @HarmonySosa

Please when you can let's tackle this issue so we can close it :)

HarmonySosa commented 1 month ago

Hi @GemmaTuron , So far I have tested the model in Ersilia with various individual molecules and it seems good; I have gotten probabilities that match what I get when I run the online model from the research paper: https://dictrank.serve.scilifelab.se/Submit_single_SMILES

GemmaTuron commented 1 month ago

wonderful! thanks @HarmonySosa !

GemmaTuron commented 1 week ago

I'll close this as completed and mark the model as done