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!

[samuelmaina]

HI @GemmaTuron , I am able to fetch the model successfully . eos85a3_cli_fetch.log

Fails to predict for one molecule.

(ersilia) @samuelmaina ➜ /workspaces $ ersilia -v run -i "O=C(O[C@H]1C[N+]2(CCCOc3ccccc3)CCC1CC2)C(O)(c1cccs1)c1cccs1"
19:48:14 | DEBUG    | Getting session from /home/codespace/eos/session.json
19:48:14 | DEBUG    | Getting session from /home/codespace/eos/session.json
19:48:14 | WARNING  | Lake manager 'isaura' is not installed! We strongly recommend installing it to store calculations persistently
19:48:14 | ERROR    | Isaura is not installed! Calculations will be done without storing and reading from the lake, unfortunately.
19:48:15 | DEBUG    | Is fetched: True
19:48:15 | DEBUG    | Schema available in /home/codespace/eos/dest/eos85a3/api_schema.json
19:48:15 | DEBUG    | Setting AutoService for eos85a3
19:48:15 | INFO     | Service class provided
19:48:19 | DEBUG    | No file splitting necessary!
19:48:20 | DEBUG    | Reading card from eos85a3
19:48:20 | DEBUG    | Reading shape from eos85a3
19:48:20 | DEBUG    | Input Shape: Single
19:48:20 | DEBUG    | Input type is: compound
19:48:20 | DEBUG    | Input shape is: Single
19:48:20 | DEBUG    | Importing module: .types.compound
19:48:20 | DEBUG    | Checking RDKIT and other requirements necessary for compound inputs
19:48:22 | DEBUG    | InputShapeSingle shape: Single
19:48:22 | DEBUG    | API eos85a3:run initialized at URL http://127.0.0.1:54549
19:48:22 | DEBUG    | Schema available in /home/codespace/eos/dest/eos85a3/api_schema.json
19:48:22 | DEBUG    | No file splitting necessary!
19:48:22 | DEBUG    | Reading card from eos85a3
19:48:22 | DEBUG    | Reading shape from eos85a3
19:48:22 | DEBUG    | Input Shape: Single
19:48:22 | DEBUG    | Input type is: compound
19:48:22 | DEBUG    | Input shape is: Single
19:48:22 | DEBUG    | Importing module: .types.compound
19:48:22 | DEBUG    | Checking RDKIT and other requirements necessary for compound inputs
19:48:22 | DEBUG    | InputShapeSingle shape: Single
19:48:22 | DEBUG    | API eos85a3:run initialized at URL http://127.0.0.1:54549
19:48:22 | DEBUG    | Schema available in /home/codespace/eos/dest/eos85a3/api_schema.json
19:48:22 | DEBUG    | Posting to run
19:48:22 | DEBUG    | Batch size 100
19:48:22 | DEBUG    | Schema available in /home/codespace/eos/dest/eos85a3/api_schema.json
19:48:53 | DEBUG    | Status code: 500
19:48:53 | ERROR    | Status Code: 500
19:48:53 | DEBUG    | Schema available in /home/codespace/eos/dest/eos85a3/api_schema.json
19:48:53 | DEBUG    | Done with unique posting
{
    "input": {
        "key": "ASMXXROZKSBQIH-VITNCHFBSA-N",
        "input": "O=C(O[C@H]1C[N+]2(CCCOc3ccccc3)CCC1CC2)C(O)(c1cccs1)c1cccs1",
        "text": "O=C(O[C@H]1C[N+]2(CCCOc3ccccc3)CCC1CC2)C(O)(c1cccs1)c1cccs1"
    },
    "output": {
        "outcomes": [
            null
        ]
    }
}

but I am getting this all the values as nulls when I run the code when I try to predict with .csv file . I had run 50 smiles and all produced 500 status. Here is the prediction for the first 10, all the vals are null

eos85a3_cli_eml_pred.csv

[GemmaTuron]

Hi @samuelmaina Which system are you using, are you working on your Mac? @febielin did you encounter anything similar at model refactoring?

[samuelmaina]

Hi @samuelmaina Which system are you using, are you working on your Mac? @febielin did you encounter anything similar at model refactoring?

Hello @GemmaTuron , I am using CodeSpaces on mac.

[febielin]

Hi @samuelmaina,

Thanks for bringing this to my attention. I am testing this out on codespaces on my mac too, just to see if I can replicate this error. So far, I've fetched, served, and run single input and compound_singles.csv input. These all seem to be working fine. Here is the log: eos85a3_fetch.log

I am currently running the eml_canonical file, so that will take sometime but I will update you when done.

[febielin]

I've been running Colab in the mean time: https://drive.google.com/file/d/1dnruy6EsR_Gn9Jmp7uHxBK55_PFBchAY/view?usp=sharing

Seems to be working well here. Have you attempted Colab yet? Please confirm that it is also working for you.

[febielin]

@samuelmaina The one molecule you mentioned above seems to work for me on CLI:

(ersilia) febielin@Febies-MacBook-Air Ersilia % ersilia run -i "O=C(O[C@H]1C[N+]2(CCCOc3ccccc3)CCC1CC2)C(O)(c1cccs1)c1cccs1"
{
    "input": {
        "key": "ASMXXROZKSBQIH-VITNCHFBSA-N",
        "input": "O=C(O[C@H]1C[N+]2(CCCOc3ccccc3)CCC1CC2)C(O)(c1cccs1)c1cccs1",
        "text": "O=C(O[C@H]1C[N+]2(CCCOc3ccccc3)CCC1CC2)C(O)(c1cccs1)c1cccs1"
    },
    "output": {
        "outcomes": [
            2.4531041177691955
        ]
    }
}

[GemmaTuron]

@samuelmaina please check that the model works in Colab and from Docker, and try again on codespaces to be sure, so we can close this issue! thanks

[samuelmaina]

Hello @GemmaTuron , Let me retest the model and see if the error persists

[GemmaTuron]

@samuelmaina When you have time please complete the task above se we can close this issue, thanks!

[samuelmaina]

Hello @GemmaTuron and @febielin , I was able to find the results in the first run. The results are consistent. eos85a3_cli_pred.csv