Amber ff19ipq is not a complete/published release

[jeremyleung521]

I just noticed that OpenMMForceFields added Amber ff19ipq in the latest release. Unfortunately ff19ipq is not complete (unpublished) and the files in the Amber main repo are work-in-progress files. I would recommend "downgrading" to ff15ipq (or ff15ipq-m, which includes ff15ipq and a set of 25 protein mimetic residues).

Suggested steps:

1. I would take one of the following: dat/leap/cmd/leaprc.mimetic.ff15ipq (ff15ipq + a set of protein mimetic residues) paper dat/leap/cmd/leaprc.fluorine.ff15ipq (ff15ipq + a set of fluorinated residues) paper

   or if you really want just the vanilla ff15ipq from 2016: dat/leap/cmd/leaprc.protein.ff15ipq (paper-cited for ff19ipq currently)

2. I would also suggest adding in the SPC/E_b water model. The current water in openMMFF repo (from 8 years ago) is the plain SPC/E. SPC/E_b has modified eq bond length (iirc) and is what ff15ipq is parameterized for/with. paper dat/leap/cmd/leaprc.water.spceb

Source: I'm a Amber ff15ipq-m developer/author :) Also if you don't believe me, check https://ambermd.org/AmberModels_proteins.php.

TLDR: ff19ipq doesn't really exist, bar from those WIP files, and were not worked on for a long time.

[epretti]

Thanks for this information! Sorry for not replying sooner as this must have gotten lost in my inbox.

We already convert ff15ipq; we could consider adding these other residues. Do leaprc.mimetic.ff15ipq and leaprc.fluorine.ff15ipq just contain extra residues and parameters for them that could be added onto leaprc.protein.ff15ipq, or do they contain parameters incompatible with the original ff15ipq? In other words, could we theoretically add them onto the ff15ipq that we are currently converting, or would each of these need to be converted as separate force fields that could only be used one at a time? We could also look into adding SPC/E-b. I will add these to the growing list of cleanups and improvements I want to do to the Amber force field suite conversion in #393. I had been working on this but it was preempted by other things and I haven't gotten back to it.

As for the fact that ff19ipq is not a complete force field, we could remove it; it would break the workflow of anyone who started using it, but perhaps they shouldn't use it if it isn't meant for use. This issue in general highlights one of my gripes with the Amber force field distributions, which is that the force fields come as a large collection of leaprc files with no reliable indication of which ones are current, experimental, outdated, or mutually incompatible with each other. We added ff19ipq since we wanted to update our force field distribution to include what appeared to be the latest in each "series" of Amber force fields, but apparently this wasn't one of them despite being distributed in the latest AmberTools releases.

If you have any say in this as a force field developer, it might be worth trying to get whoever's in charge of AmberTools releases / documentation to remove ff19ipq or better advertise that it shouldn't be used if that's really the case. I can find two instances in the literature [1] [2] of people using and citing (or trying to cite) ff19ipq. These papers were received before we started including it with openmmforcefields, which means they must have found it as part of AmberTools and incorrectly assumed like me that it was a new successor of ff15ipq.

[jeremyleung521]

Thanks for this information! Sorry for not replying sooner as this must have gotten lost in my inbox.

We already convert ff15ipq; we could consider adding these other residues. Do leaprc.mimetic.ff15ipq and leaprc.fluorine.ff15ipq just contain extra residues and parameters for them that could be added onto leaprc.protein.ff15ipq, or do they contain parameters incompatible with the original ff15ipq? In other words, could we theoretically add them onto the ff15ipq that we are currently converting, or would each of these need to be converted as separate force fields that could only be used one at a time? We could also look into adding SPC/E-b. I will add these to the growing list of cleanups and improvements I want to do to the Amber force field suite conversion in #393. I had been working on this but it was preempted by other things and I haven't gotten back to it.

As for the fact that ff19ipq is not a complete force field, we could remove it; it would break the workflow of anyone who started using it, but perhaps they shouldn't use it if it isn't meant for use. This issue in general highlights one of my gripes with the Amber force field distributions, which is that the force fields come as a large collection of leaprc files with no reliable indication of which ones are current, experimental, outdated, or mutually incompatible with each other. We added ff19ipq since we wanted to update our force field distribution to include what appeared to be the latest in each "series" of Amber force fields, but apparently this wasn't one of them despite being distributed in the latest AmberTools releases.

If you have any say in this as a force field developer, it might be worth trying to get whoever's in charge of AmberTools releases / documentation to remove ff19ipq or better advertise that it shouldn't be used if that's really the case. I can find two instances in the literature [1] [2] of people using and citing (or trying to cite) ff19ipq. These papers were received before we started including it with openmmforcefields, which means they must have found it as part of AmberTools and incorrectly assumed like me that it was a new successor of ff15ipq.

Thanks for replying @epretti ! No worries, it's a busy time of the year.

1. ff15ipq-m and ff15ipq (with fluorine) should be compatible together (though we haven't tried and don't think people would be doing both protein mimetics and fluorinated residues at the same time). They were just developed independently and thus distributed separately but they shouldn't interfere with each other as far as I can tell. They both use the common ff15ipq base, which hasn't changed since release in 2015.
2. I understand the frustration (and understand why people might've thought ff19ipq is the most recent). I also started the process of requesting removal of ff19ipq from the main Amber repo a few weeks back. It wasn't really a priority until I saw those papers using ff19ipq, which worries me a little.

[epretti]

OK, in that case, we'll stop distributing ff19ipq. Out of curiosity, do you know who is/was working on developing it? We'd want to know who to reach out to so we can support it in the future if it is ever completed (unless its development has been abandonded).

[jeremyleung521]

OK, in that case, we'll stop distributing ff19ipq. Out of curiosity, do you know who is/was working on developing it? We'd want to know who to reach out to so we can support it in the future if it is ever completed (unless its development has been abandonded).

It's abandoned. Dave Cerutti was working on it, but had moved on from academia to industry. The files were merged in around his transition period, probably by mistake. Even if the project gets picked up somehow, I highly doubt it'll be called ff19ipq (considering that's 6 years ago).

[epretti]

ff19ipq has been removed.

Since this issue also contains a suggestion to add SPC/Eb as a supported water model, I'll leave it open for now as a feature request.