Hamiltonian Replica Exchange and CMAP

[jus...@uchicago.edu]

Hi all,

I am attempting to run Hamiltonian Replica Exchange (H-REMD) on a system with intrinsically disordered proteins (IDPs). Given this system, my choice of force field is quite important, with most force fields being quite bad for IDPs. Those I've found that seem suited for this are CHARMM36/36m, AMBER ff14IDP, and some other AMBER variants.

Unfortunately, the topology scaling mechanism provided with PLUMED states that it improperly operates on force fields with CHARMM CMAP. However, I believe that the newer, IDP-friendly AMBER variants also implement CMAP. So--

Q1: Can this implementation of topology scaling be used with AMBER given its CMAP implementation?

If so, that's great. If not, I've got a few other questions.

I may be ignoring useful force fields for this purpose. So--

Q2:  Does anyone recommend other force fields for IDPs, beyond CHARMM and AMBER? 

If AMBER does not work here and no IDP-friendly, PLUMED-friendly force field exists, it may be necessary to use a different implementation:

Q3: Does a decent implementation of H-REMD exist outside of PLUMED?

Thanks!

Jacob

[Giovanni Bussi]

Hi,

it depends on what you need to scale. For solute tempering (whole solute scaled) there should be no problem. To be precise, there are two things to do:

- prepare scaled topologies (e.g. with "plumed partial_tempering"). I think this is not capable to scale cmap terms. Search for "cmap hrex" on this forum to find some possible solution. You might even use the  "plumed partial_tempering" and then scale by hand the cmap terms with a script. If I remember correctly the gromacs format, this would be very problematic for "partial" tempering schemes, but solute tempering will work fine.

- run with mdrun -hrex. This will work fine whatever change you do to the topologies (provided masses are not changed)

Giovanni

--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users+unsubscribe@googlegroups.com.
To post to this group, send email to plumed...@googlegroups.com.
Visit this group at https://groups.google.com/group/plumed-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/4fa58316-2581-4232-a018-5dfd02d19aa7%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[dbu...@iastate.edu]

Hi, 

Can you tell me what specifically the "plumed partial_tempering" is applying scaling factors to?  I'm using a Charmm forcefield and want to go back and edit the terms that got left out.  I see the charge has been scaled for my "hot" atoms but the commented ";qtot" at the end of the moleculetype line is not changed between systems of different scaling factors.  

What other values should be lambda scaled?  I see there is a cp column in the dihedraltypes section - is that a force constant that should be scaled?

Thank you,

Dan

On Wednesday, September 13, 2017 at 2:53:17 AM UTC-5, Giovanni Bussi wrote:

Hi,

it depends on what you need to scale. For solute tempering (whole solute scaled) there should be no problem. To be precise, there are two things to do:

- prepare scaled topologies (e.g. with "plumed partial_tempering"). I think this is not capable to scale cmap terms. Search for "cmap hrex" on this forum to find some possible solution. You might even use the  "plumed partial_tempering" and then scale by hand the cmap terms with a script. If I remember correctly the gromacs format, this would be very problematic for "partial" tempering schemes, but solute tempering will work fine.

- run with mdrun -hrex. This will work fine whatever change you do to the topologies (provided masses are not changed)

Giovanni

On Tue, Sep 12, 2017 at 8:34 PM, <jus...@uchicago.edu> wrote:

Hi all,

I am attempting to run Hamiltonian Replica Exchange (H-REMD) on a system with intrinsically disordered proteins (IDPs). Given this system, my choice of force field is quite important, with most force fields being quite bad for IDPs. Those I've found that seem suited for this are CHARMM36/36m, AMBER ff14IDP, and some other AMBER variants.

Unfortunately, the topology scaling mechanism provided with PLUMED states that it improperly operates on force fields with CHARMM CMAP. However, I believe that the newer, IDP-friendly AMBER variants also implement CMAP. So--

Q1: Can this implementation of topology scaling be used with AMBER given its CMAP implementation?

If so, that's great. If not, I've got a few other questions.

I may be ignoring useful force fields for this purpose. So--

Q2:  Does anyone recommend other force fields for IDPs, beyond CHARMM and AMBER? 

If AMBER does not work here and no IDP-friendly, PLUMED-friendly force field exists, it may be necessary to use a different implementation:

Q3: Does a decent implementation of H-REMD exist outside of PLUMED?

Thanks!

Jacob

--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to plumed...@googlegroups.com.

[Giovanni Bussi]

Hi,

the comment after ";" is not touched since, well, it is a comment. I am not even sure the syntax is well defined in gromacs.

The script scales:

- charges of hot atoms using sqrt(lambda)

- LJ epsilon of new atom types with additional underscores using lambda. hot atoms are then assigned to those types

- All dihedral angles where both atoms 1 and 4 are hot are scaled using lambda

- All dihedral angles where only one between atoms 1 and 4 is hot are scaled using sqrt(lambda)

In case you find something different then please report it as a bug.

For cmaps, see my previous mail.

Giovanni

To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/41a4d7bd-afe2-4e7f-8e0b-08de379180e6%40googlegroups.com.

[Daniel Burns]

Okay, thank you.

My question about the commented total charge is, why does the qtot for the "hot" atom not change between different scaling factors?

I've tested scaled and unscaled systems with charmm36 forcefield with cmap terms. With a cursory analysis, they seem to be behaving as desired.  I am not scaling the whole protein, just a region so I've avoided using the awk script to correct the cmap terms.

 If I want to generate an ensemble of conformations and analyze their free energy profiles, do you know what the consequence would be of leaving the cmap values unchanged?

Thank you,

Dan

You received this message because you are subscribed to a topic in the Google Groups "PLUMED users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/plumed-users/lo-Sb8aUepY/unsubscribe.
To unsubscribe from this group and all its topics, send an email to plumed-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/CAPLm8ZKg3XLXTorxM4iB4SFCioJ9uZ4G5coFiZzXeHzJy8pVWA%40mail.gmail.com.

[Giovanni Bussi]

On Tue, Feb 11, 2020 at 7:09 PM Daniel Burns <dbu...@iastate.edu> wrote:

Okay, thank you.

My question about the commented total charge is, why does the qtot for the "hot" atom not change between different scaling factors?

As I wrote, because that's a comment only. It is not read by gromacs, so no need to change it.

 

I've tested scaled and unscaled systems with charmm36 forcefield with cmap terms. With a cursory analysis, they seem to be behaving as desired.  I am not scaling the whole protein, just a region so I've avoided using the awk script to correct the cmap terms.

 If I want to generate an ensemble of conformations and analyze their free energy profiles, do you know what the consequence would be of leaving the cmap values unchanged?

You wrote above that you can scale cmap correctly so why are you interested in this?

Anyway, if you don't scale the cmap you can expect that sampling on those dihedral angles affected by the cmaps could be suboptimal. Results would be correct anyway if the simulation is long enough

Giovanni

 

To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/CAGdmtvCLiUDozhXEOQgRVohuGQUD65B%3DcHkk6_DhHZy%3DBO-53g%40mail.gmail.com.