From: JC Gumbart (gumbart_at_ks.uiuc.edu)
Date: Thu Jul 31 2014 - 16:24:06 CDT
This is really good to know. Somehow that detail escaped me in the past.
> The CHARMM community is a bit divided over whether it's justified to use a 2fs time step.
This is funny, as I’m reading now about whether even a 4-fs time step is acceptable when using hydrogen mass repartitioning. My guess is you would say no? I read the namd-l discussion from a couple years ago in which Axel chose not to give an opinion then.
JC
On Jul 31, 2014, at 3:03 PM, Kenno Vanommeslaeghe <kvanomme_at_rx.umaryland.edu> wrote:
> It's slightly more complicated than that. Strictly spoken, even with SHAKE, a 2fs time step is too long if you want sensitive bulk phase properties to converge (as a function of the time step) within a few %. In line with this, the bulk phase simulations we perform to optimize the nonbonded parameters are done with SHAKE on all covalent bonds involving hydrogens (the equivalent to "rigidBonds all" in NAMD) *and* a 1fs time step, so that would be the most rigorously correct use of the force field. The CHARMM community is a bit divided over whether it's justified to use a 2fs time step. The prevalent opinion on our lab is that it almost never significantly alters the properties of interest in typical biomolecular simulations, so we use 2fs for most of our biomolecular work. Other labs default to 1fs. However, we all agree(*) on two things:
> - for very fine quantitative work (such as force field parametrization but also certain detailed studies of lipid membrane properties), 1fs is the standard
> - no matter whether using 1fs or 2fs, "SHAKE BONH" / "rigidBonds all" should always be applied (If the methodology/software allows for it. The BOND parameters provide a workaround/safety net if it doesn't, but at the price of some accuracy with respect to the parametrization.)
>
> (*) famous last words
>
> On 07/30/2014 11:27 AM, JC Gumbart wrote:
>>> Though if you're using the CHARMM force field, it's actually recommended to instead use
>>> rigidBonds all
>>> because that's what the force field is parametrized for.
>>
>> I thought this was only the case when using a 2-fs time step?
>>
>>
>> On Jul 30, 2014, at 12:21 AM, Kenno Vanommeslaeghe <kvanomme_at_rx.umaryland.edu> wrote:
>>
>>> On 07/29/2014 06:59 PM, Hadi wrote:
>>>> Below you can find the parameters of the modified TIP3P model (CGenFF:
>>>> Parameters for the Charmm General Force Field). This model is often used
>>>> in NAMD.
>>>
>>> (1) Small detail: the only thing this has to do with CGenFF is that it is the water model on which *all* the all-atom additive CHARMM force fields are based, the first one of which predates CGenFF by roughly 15 years. The confusion likely arises from the fact that the NAMD version of the CHARMM water stream file can currently be found on my CGenFF page. Instead, we probably should put it on the main CHARMM force field page alongside the CHARMM version.
>>>
>>> (2) It is my understanding that the parameters you copy-pasted into your e-mail are just "reasonable values" for in case an MD engine or methodology is used that does not support rigid water. They won't give the exact same properties as the true rigid TIP3P model. NAMD, however, does support rigid water; assuming the psf is generated correctly, all one need to do is put the following line in the NAMD input:
>>> rigidBonds water
>>> Though if you're using the CHARMM force field, it's actually recommended to instead use
>>> rigidBonds all
>>> because that's what the force field is parametrized for.
>>
This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:22:41 CST