Question about reproducing MutationalPPI results

[StevenD07]

Hi, thank you for sharing this repository — great work!

I have a question regarding the reproduction of the MutationalPPI results.

I tried to reproduce the reported results using the following commands:

python downstream/GeneralPPI/embeddings_mint.py --task MutationalPPI --sep_chains
python finetune_general.py --task MutationalPPI --model mint --sep

However, the performance I obtained is significantly different from what is reported in the paper.

I was wondering whether the correct task setting should instead be:

python downstream/GeneralPPI/embeddings_mint.py --task MutationalPPI_cs --model_name mint --sep_chains

However, when following downstream/GeneralPPI/mutational-ppi/prepare_data.ipynb, it does not appear to generate data for the MutationalPPI_cs task.

Could you please clarify whether MutationalPPI or MutationalPPI_cs should be used to reproduce the reported results?

Thank you very much for your help.

[VarunUllanat]

Hello, you are right in using the --task MutationalPPI argument. However, I see that I ran this using two additional arguments, so python finetune_general.py --task MutationalPPI --model mint --normalize --use_mlp_for_cv --sep.

Let me know if this works!