DeepPurpose
DeepPurpose copied to clipboard
kexinhuang12345
Creating Visual Changes to Training Process Using tqdm and Fixing Graph-Based Models Issue With Better Memory Managment
@kexinhuang12345 Hi, Can you please review the pull request and merge it? Best, Mehdi
Hi Mehdi, thank you for the PR! Did you do any test to see if it works across all the case studies? There are too many changes so i will need to spend some time to read through. It would be great if you could run all of these case studies from 1a-e and 2a-b, and see if they work as normal:
https://github.com/kexinhuang12345/DeepPurpose#case-study-1a-a-framework-for-drug-target-interaction-prediction-with-less-than-10-lines-of-codes
If you need a processed dataset for c-e, please let me know!
Hi Mehdi, thank you for the PR! Did you do any test to see if it works across all the case studies? There are too many changes so i will need to spend some time to read through. It would be great if you could run all of these case studies from 1a-e and 2a-b, and see if they work as normal:
https://github.com/kexinhuang12345/DeepPurpose#case-study-1a-a-framework-for-drug-target-interaction-prediction-with-less-than-10-lines-of-codes
If you need a processed dataset for c-e, please let me know!
sure but the main version does not work either!
wait, what does not work? can you give a script?
The first case study does not work. Exact same script used in the example!
I see indeed. The new bindingDB dataset has some big molecule and the MPNN does not support that. I update it to 'CNN' you can test it out. Ideally, it would be great to test DGL based encoding since you made changes on that.