Open mmagithub opened 1 year ago
Hey @mmagithub, thanks for your interest. Yes, the old AIMNet model has an SMD solvent effect baked in. You can hack Auto3D and use it if you wish, but it works only for neutral molecules. However, we are very close (~month-ish) to a new AIMNet2 model that should work for different charged/protonation states. Please note that we are not doing any pKa calculations yet.
That would be wonderful, thanks for your answer.
Hey, @isayev / @LiuCMU, is this enhancement already available? Would be keen to try it out!
Hey @mmagithub, thanks for your interest. Yes, the old AIMNet model has an SMD solvent effect baked in. You can hack Auto3D and use it if you wish, but it works only for neutral molecules. However, we are very close (~month-ish) to a new AIMNet2 model that should work for different charged/protonation states. Please note that we are not doing any pKa calculations yet.
Hi @OleinikovasV, the default model in Auto3D is AIMNet2. If you specify optimizing_engine="AIMNET"
, it actually uses AIMNet2. The old AIMNet model has been deprecated since Auto3D 2.2.1, and every call to "AIMNET" refers to the AIMNet2 model.
Is it possible to do conformer generation with and without a solvent model (gas and aqueous phase) in Auto3D? It sounds like only gas phase calculations are currently available, but that the older optimization engine has a solvation model.
Greetings and good day,
I am wondering if auto3d can enumerate potential protonation states, same way as tautomeric states are being enumerated, I could not find anything relevant in the auto3d paper.
I am also wondering if solvent effect has been implemented yet (SMD or similar), as according to my understanding from the paper, it was an 'under development' functionality. SMD is already implemented in AIMNet, which auto3d uses by default as an optimisation engine.
Thanks, Marawan