New model ready for testing!

[github-actions[bot]]

This model is ready for testing. If you are assigned to this issue, please try it out using the CLI, Google Colab and DockerHub and let us know if it works!

[GemmaTuron]

Hi @febielin and @pittmanriley ! Can you check that this model works?

Thanks :)

[febielin]

Hi @GemmaTuron,

I have tested this model on CLI, Google Colab, and DockerHub, and it appears functional.

My process is testing the model on a string format input of a molecule, as well as a smiles file input, both of which have been successful.

I am still figuring out the process of testing models, so please let me know if there is anything else that I could be doing to make this testing process more robust.

FYI, Riley is still experiencing issues fetching the model on all three platforms, but I was able to share my screen and show him my process. He can verify that it is working on the 3 platforms.

[febielin]

Forgot to mention, I also created a file with 20 compounds that I picked at random from [ersilia-model-hub-maintained-inputs] and that ran as well.

Please let me know how I can make my testing of the models more comprehensive.

[GemmaTuron]

Thanks @febielin !

We aim to put together a more comprehensive documentation of model testing (see discussions) but haven't got to it yet. Meanwhile, testing should be:

• As you did, test on the CLI with a .csv input and .csv output
• Testing using the Google Colab implementation

[GemmaTuron]

@pittmanriley Does this model work on your system?

[pittmanriley]

@GemmaTuron, I was unable to get this model to work on my system. Is there anything further you’d like me to do to try with this model?

[GemmaTuron]

Hi @pittmanriley

We need to make sure the models work in all systems, why this one does not work at this moment? can you please share more?

[pittmanriley]

Hi @GemmaTuron, Miquel just helped me get this model working on my computer. Since I am using a Mac with an M1 chip, I have issues fetching models that are only available in AMD64. However, I was able to get it to work with the following commands:

• docker pull ersiliaos/eos74bo:latest --platform linux/amd64
• ersilia -v fetch eos74bo

The Ersilia codebase is now updated with a new exception that automatically does this when we fetch. I will discuss this more in the meeting tomorrow when I present.

[pittmanriley]

Here are my results when I tested it. It seems to be working:

ersilia run -i compound_list.csv { "input": { "key": "LUHMMHZLDLBAKX-UHFFFAOYSA-N", "input": "CC1C2C(CC3(C=CC(=O)C(=C3C2OC1=O)C)C)O", "text": "CC1C2C(CC3(C=CC(=O)C(=C3C2OC1=O)C)C)O" }, "output": { "outcome": [ 0.008 ] } } { "input": { "key": "QRXWMOHMRWLFEY-UHFFFAOYSA-N", "input": "C1=CN=CC=C1C(=O)NN", "text": "C1=CN=CC=C1C(=O)NN" }, "output": { "outcome": [ 0.983 ] } } { "input": { "key": "SGOIRFVFHAKUTI-UHFFFAOYSA-N", "input": "CC(CN1C=NC2=C(N=CN=C21)N)OCP(=O)(O)O", "text": "CC(CN1C=NC2=C(N=CN=C21)N)OCP(=O)(O)O" }, "output": { "outcome": [ 0.009 ] } }

When I tested in Google Colab, it also successfully ran. Here's my output file, tested on eml_canonical.csv: eos74bo_output.csv

[GemmaTuron]

Thanks @pittmanriley and @febielin I'll close this as completed. For next times, is best to test the output from the CLI to a .csv file as well, to ensure it is not failing at parsing the .json to csv