Re: [AMBER] Help needed on MMPBSA for a system with different protonation states in free & bound forms, using multiple trajectory protocol

From: Jason Swails <>
Date: Sat, 16 May 2015 12:59:21 -0400

On Sat, May 16, 2015 at 7:18 AM, Shailesh Pandey <>

> Dear Amber Users,
> I am trying to use with multiple trajectory protocol for
> estimation of binding free energy of a ligand to its receptor. But in
> binding site of receptor(a protein kinase) a histidine is predicted by
> PROPKA3.0 to have different protonation state in free and bound states with
> apo and holo crystal structures respectively. Due to this fails
> to detect receptor and ligand residues even if I provide it explicity in
> the file.
> content of is as below
> Input file for running PB and GB
> &general
> startframe = 1, endframe = 999999, interval = 20, verbose=1,
> entropy = 0,
> /
> &gb
> igb=2, saltcon=0.001,
> /
> &pb
> istrng=0.001, radiopt=0, inp=1, scale = 2.0, fillratio = 4.0,
> /
> I used command as below to invoke
> -cp ../02.leap/com.gas.leap.prmtop \
> -rp ../02.leap/rec.gas.leap.prmtop \
> -lp ../02.leap/lig.gas.leap.prmtop \
> -y ../04.ptraj/com.stripfit.mdcrd \
> -yr ../04.ptraj/rec.stripfit.mdcrd \
> -yl ../04.ptraj/lig.strip.mdcrd
> File "/home/shailesh/soft/amber14/bin/", line 94, in <module>
> app.loadcheck_prmtops()
> File "/home/shailesh/soft/amber14/bin/MMPBSA_mods/", line 608, in
> loadcheck_prmtops
> self.normal_system.Map(INPUT['receptor_mask'], INPUT['ligand_mask'])
> File "/home/shailesh/soft/amber14/bin/MMPBSA_mods/", line
> 242, in Map
> "this is not the case.")
> Checking in the, line 242, I found that, this section of code
> expects, complex residue sequence to be same as (receptor + ligand)
> sequence, but if protonation states are different for any of residue in
> free & bound forms, then this condition will never be satisfied, because
> diiferent residue names are used for different protonation states.
> So, I made a copy of the original receptor prmtop as say
> temp_receptor.prmtop and edited its RESIDUE_LABEL section to match
> RESIDUE_LABEL section of complex.prmtop
> Now trying MMPBSA with temp_receptor.prmtop, It passes this section but
> gets caught at charge-cosistency check and error is as below:
> File "/share/apps/amber14/bin/", line 94, in <module>
> app.loadcheck_prmtops()
> File "/share/apps/amber14/bin/MMPBSA_mods/", line 609, in
> loadcheck_prmtops
> self.normal_system.CheckConsistency()
> File "/share/apps/amber14/bin/MMPBSA_mods/", line 924, in
> CheckConsistency
> (i + 1))
> PrmtopError: Inconsistent charge definition for atom 80!
> As it is obvious to have different charges for residue's atoms with
> different protonation state in free & bound form.
> Has anyone faced such things earlier, what can be ways to solve it.
> or
> Is there a way to skip prmtop consistency check ??

​You can comment it out if you want, but there is a good reason it's there.

> ​​
> Any help/suggestion in this regard is highly appreciated.

​What you are trying to do is complicated. If your charges are different
in your bound and unbound states, any Delta G you get from will
be meaningless. In the case of a protonation event, the thermodynamic
cycle upon which end-state free energy methods are built do not take into
account protonation state changes (which involve, among other things,
charge redistribution, H-O bond breaking, H+ desolvation, and other quantum
effects). The force field doesn't get those values anywhere *near*
correct, and classical approaches to model this behavior (e.g., constant pH
MD) constructs a thermodynamic cycle so they can use experimental values to
calibrate their treatment of deprotonation events (effectively reducing the
problem to one of a "change of a change" of free energy, which is easier to
compute than a simple free energy change upon protonation).​

Your problem is not impossible, but it's more complicated than
was intended to deal with. Just removing the checks will let the
simulation run (maybe), but it won't make the results meaningful. There is
no tool yet for doing this kind of study (although I've given thought as to
how it could be done), so you will likely need to run this binding
calculation by hand after devising the correct thermodynamic cycle for it.


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Sat May 16 2015 - 10:00:03 PDT
Custom Search