Hello. I am an artificial intelligence engineer and I have been working with a drug discovery expert. We have found three new ligands for the SARS-CoV-2 Mpro through deep learning via autoencoders. The problem is, we didn’t design them for any fragment IDs. We just took existing drugs, generated new ones from those existing drugs via autoencoders, docked them to come up with binding energies, and ranked them. More details in the paper: https://chemrxiv.org/articles/On_the_Generation_of_Novel_Ligands_for_SARS-CoV-2_Protease_and_ACE2_Receptor_via_Constrained_Graph_Variational_Autoencoders/12011157
Please advise on how to proceed and submit. Thank you so much!
Hi, sorry about that, please just pick a the first fragment and in “Notes” please note that you do not actually use any of the fragments
1 Like
Thank you! I have submitted them already. JAS-UNI-326
Hi, this is very interesting, thank you for sharing. I took a similar approach, but I didn’t use neural networks. Also, to create the generative model, I used the current set of submissions on this website (which was ~2,000 molecules as of Mar. 27, 2020). These are my submissions:
https://covid.postera.ai/covid/submissions/a1be81af-ddd9-4c01-8546-0400b9ee4a4e
https://covid.postera.ai/covid/submissions/2c46affe-719d-43fe-ac03-722bb8f41f87
That’s nice to know. When other people have similar methodologies, it validates their methods more. Cheers!