search

TinkerTools / tinker

Tinker: Software Tools for Molecular Design
https://dasher.wustl.edu/tinker/
Other
130 stars 61 forks source link

poledit frame assignment error #71

Closed misterbrandonwalker closed 3 years ago

misterbrandonwalker commented 3 years ago

9_8_Index_0-peditin.txt 9_8_Index_0.sdf.txt 9_8_Index_0.gdmaout.txt

Hello, I noticed the hydrogens on the benzene in this molecule were assigned multipole frames of ions. Poltype relies on poledit defaults now instead of making our own in the input peditin.txt file.

Thanks for your time,

misterbrandonwalker commented 3 years ago

9_8_Index_0.key.txt

Output key file

misterbrandonwalker commented 3 years ago

9_8_Index_0.xyz.txt Okay it would seem that sometimes 3 cycles of OPT are not long enough sometimes, we have H bonds of 1.2,

jayponder commented 3 years ago

Hi Brandon, Since POLEDIT works only from the DMA output, it has to decide when atoms are close enough to be "bonded". The tolerances currently used are based on scaled covalent radii for each element. At present, for example, a carbon and hydrogen have to be 1.2615 Ang or closer to be considered bonded. If you have C-H bonds in a molecule that are longer than that, then I'd say your structure is not very good, and you probably can't trust the DMA values from QM on such a bad structure!

If you want, I can perhaps loosen the tolerances a bit more, but I think the current values are probably very good. If you make the distance tolerance much bigger then you start to pick up inappropriate things as bonds, which causes POLEDIT problems as well. I do not know of any C-H bond length in a real organic molecule that is greater than 1.26 Ang (in fact I don't know any greater than about 1.15 Ang).

If you are OK with the current settings, then I'll close this. You guys need to be careful to do enough QM optimization to get reasonable structures. I would guess that a small number of QM optimization cycles should generally be fine, but that asusmes your input structure already has at least reasonable (ie, not "crazy") bond and angle values. One idea would be to take a crude structure from, say, some arbitrary SDF file that could have any crazy stuff in it, and run it through a quick optimization (seconds) in Tinker using the very simple force field in "tiry.prm" from the Tinker distribution. That simple force field was more or less designed for use in "regularizing" bonds and angles in arbitrary structures.

misterbrandonwalker commented 3 years ago

Yes, sounds very reasonable, thank you! We will attend to this optimization issue very soon.

From: Jay Ponder notifications@github.com Sent: Saturday, November 14, 2020 12:26:32 PM To: TinkerTools/tinker tinker@noreply.github.com Cc: Brandon Walker bdw2292@utexas.edu; Author author@noreply.github.com Subject: Re: [TinkerTools/tinker] poledit frame assignment error (#71)

Hi Brandon, Since POLEDIT works only from the DMA output, it has to decide when atoms are close enough to be "bonded". The tolerances currently used are based on scaled covalent radii for each element. At present, for example, a carbon and hydrogen have to be 1.2615 Ang or closer to be considered bonded. If you have C-H bonds in a molecule that are longer than that, then I'd say your structure is not very good, and you probably can't trust the DMA values from QM on such a bad structure!

If you want, I can perhaps loosen the tolerances a bit more, but I think the current values are probably very good. If you make the distance tolerance much bigger then you start to pick up inappropriate things as bonds, which causes POLEDIT problems as well. I do not know of any C-H bond length in a real organic molecule that is greater than 1.26 Ang.

If you are OK with the current settings, then I'll close this. You guys need to be careful to do enough QM optimization to get reasonable structures. I would guess that a small number of QM optimization cycles should generally be fine, but that asusmes your input structure already has at least reasonable (ie, not "crazy") bond and angle values.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/TinkerTools/tinker/issues/71#issuecomment-727246888, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AKNB26N7KSPDKNTUMAH7RUTSP3DVRANCNFSM4TSF7PVQ.