search

DeepGraphLearning / ProtST

[ICML-23 ORAL] ProtST: Multi-Modality Learning of Protein Sequences and Biomedical Texts
Apache License 2.0
87 stars 7 forks source link

How to run prediction of protein function #6

Open CryoSky opened 1 year ago

CryoSky commented 1 year ago

Hello,

Thank you again for your help in my previous issue. I can finally run the ProtST task on my cluster using one GPU. However, I need clarification about the command that you provided.

I want to predict a given protein sequence's function (GO terms). After running the following command, what do you think I should do to invoke the trained model and run the prediction? I appreciate your help in advance.

python4proc /script/run_downstream.py --config ./config/downstream_task/PretrainESM2/annotation_tune.yaml --checkpoint /work/cms16/sj52/scratch/esm-model-weights/esm2_t33_650M_UR50D.pt --dataset GeneOntology --branch BP

KatarinaYuan commented 1 year ago

Hi, the running command should be correct. And to use the ProtST-enhanced model, we just need to modify the augment in "--checkpoint" to the storage path of that ProtST-enhanced model.

As I commented in the previous issue (https://github.com/DeepGraphLearning/ProtST/issues/5#issuecomment-1676645310). I'm now downloading ProtST-ESM2 checkpoint onto my current working cluster to test if there is any problem. I would come back to you once it's done.

KatarinaYuan commented 1 year ago

Please check https://github.com/DeepGraphLearning/ProtST/issues/5#issuecomment-1677753219