[gmx-users] Glutamate to Alanine Mutation

Sai Kumar Ramadugu sramadugu at gmail.com
Wed Sep 26 21:31:54 CEST 2012


Dear Prof Mobley & Gromacs Community,

I have done Glutamate to Alanine mutation using TI approach in Gromacs. I
am using OPLS-AA force field for protein and I am using OPLS-AA for
carbohydrates developed by Bill Jorgensen's group for the ligand.

The system consists of a protein (101 aa) and ligand (disacchardie). The
way I split up the calculations are as follows:


Since my system has a glutamate which is -1 charged and the mutation is to
a neutral amino acid Alanine, I'm mutating a counter-ion, K+ at same time
to maintain the system neutrality. The reason for mutating and the
reference for doing this particular way is from the following paper by
Brittany Morgan et al:
Accurate Estimates of Free Energy Changes in Charge Mutations. J. Chem.
Theory Comput. 2010, 6, 1884–1893

The K+ ion is held at least 12 Angstorms apart so that the mutating ion
does not come close to the protein. I am using position restraints to
achieve this effect.

Step 1: Removal of charge calculation:


Here I am pasting the relevant part of the topology where I made changes
from Glu to Ala using dual topology approach.

; residue  40 GLU rtp GLU  q -1.0
   552   opls_238     40    GLU      N    191          -0.5      14.0067
opls_238     -0.5    14.0067 ; qtot 0.5
   553   opls_241     40    GLU      H    191           0.3       1.008
 opls_241      0.3     1.008 ; qtot 0.8
   554  opls_224B     40    GLU     CA    191         0.14    12.011
opls_224B    0.14   12.011 ; qtot 0.94
   555   opls_140     40    GLU     HA    191          0.06      1.008
opls_140      0.06    1.008 ; qtot 1
   556   opls_136     40    GLU     CB    192         -0.12     12.011
opls_135    -0.18    12.011 ; qtot 0.88
   557   opls_140     40    GLU    HB1    192         0.06      1.008
 opls_140     0.06    1.008 ; qtot 0.94
   558   opls_140     40    GLU    HB2    192         0.06      1.008
 opls_140     0.06    1.008 ; qtot 1
   559   opls_274     40    GLU     CG    193         -0.22    12.011
 opls_140     0.06    1.008 ; qtot 0.78
   560   opls_140     40    GLU    HG1    193         0.06     1.008
 DUM_140     0.0      1.008 ; qtot 0.84
   561   opls_140     40    GLU    HG2    193         0.06     1.008
 DUM_140     0.0      1.008 ; qtot 0.9
   562   opls_271     40    GLU     CD    194          0.7      12.011
DUM_271     0.0      12.011 ; qtot 1.6
   563   opls_272     40    GLU    OE1    194         -0.8     15.9994
DUM_272     0.0      15.9994 ; qtot 0.8
   564   opls_272     40    GLU    OE2    194         -0.8     15.9994
DUM_272     0.0      15.9994 ; qtot 0
   565   opls_235     40    GLU      C    195            0.5      12.011
 opls_235      0.5      12.011 ; qtot 0.5
   566   opls_236     40    GLU      O    195           -0.5     15.9994
opls_236     -0.5      15.9994 ; qtot 0


For the dummy atoms representing the changing C, H and O atoms I created
DUM_140, DUM_271 and DUM_272 and for coulomb calculation, I have included
the sigma and epsilon values of  2.50000e-01  1.25520e-01 for DUM_140,
  3.75000e-01  4.39320e-01 for DUM_271 and 2.96000e-01  8.78640e-01 for
DUM_272.

Since I am mutating the charge of the K+ ion, I am also mutating the K+ to
K0 and I have created a dummy atom for the same purpose DUM_408 with sigma
= 4.93463e-01 and epsilon = 1.37235e-03.
The topology section of mutating the potassium ion K+ to K0 atom is pasted
below:

[ moleculetype ]
; Name            nrexcl
KM               1

[ atoms ]
;   nr       type  resnr residue  atom   cgnr     charge       mass  typeB
   chargeB      massB
 1  opls-408      1       KM        KM        1      1.0        39.0983
DUM_408   0.0         39.0983


I am doing this calculation in presence and absence of the ligand.

Step 2: LJ transformation calculation

The relevant part of the topology for LJ transformation is pasted below:

; residue  40 GLU rtp GLU  q -1.0
   552   opls_238     40    GLU      N    191        0.0     14.0067 ;
   553   opls_241     40    GLU      H    191        0.0       1.008 ;
   554  opls_224B     40    GLU     CA    191      0.0      12.011 ;
   555   opls_140     40    GLU     HA    191       0.0       1.008 ;
   556   opls_136     40    GLU     CB    192       0.0      12.011 ;
   557   opls_140     40    GLU    HB1    192      0.0       1.008 ;
   558   opls_140     40    GLU    HB2    192      0.0       1.008 ;
   559   opls_274     40    GLU     CG    193       0.0      12.011
 opls_140     0.0      1.008 ; qtot 0.78
   560   opls_140     40    GLU    HG1    193      0.0       1.008
DUM_140    0.0       1.008 ; qtot 0.84
   561   opls_140     40    GLU    HG2    193      0.0       1.008
DUM_140    0.0       1.008 ; qtot 0.9
   562   opls_271     40    GLU     CD    194       0.0      12.011
 DUM_271    0.0      12.011 ; qtot 1.6
   563   opls_272     40    GLU    OE1    194      0.0     15.9994
DUM_272    0.0      15.9994 ; qtot 0.8
   564   opls_272     40    GLU    OE2    194      0.0     15.9994
DUM_272    0.0      15.9994 ; qtot 0
   565   opls_235     40    GLU      C    195        0.0      12.011 ;
   566   opls_236     40    GLU      O    195       0.0     15.9994 ;

Since I am only changing the side chain of Glu to side chain of Ala, I have
mutated only the part of the side chain that is different between the two
amino acids.
At the same time I am also mutating the K0 atom to a dummy atom. The
topology section of mutating the potassium atom to a dummy atom is pasted
below:

[ moleculetype ]
; Name            nrexcl
KM               1

[ atoms ]
;   nr       type  resnr residue  atom   cgnr     charge       mass  typeB
   chargeB      massB
 1  opls-408      1       KM        KM        1      0.0        39.0983
DUM_408   0.0         39.0983


I am doing this transformation in presence and absence of the ligand.

After these two steps:
For the analysis I am just using the values of dV/dl printed for every 10
steps from the simulation from 0  to 1 in lambda and integrating the dV/dl
w.r.t. lambda.

Step 1 charge in presence of Ligand = 790.109 kJ/mol
Step 2 vdw in presence of Ligand     = -29.324 kJ/mol

The sum of two steps in presence of ligand = 760.785 kJ/mol

Step1 charge in absence of Ligand = 787.33 kJ/mol
Step 2 vdw in absence of Ligand    = -21.8127 kJ/mol

The sum of two steps in absence of ligand = 765.517 kJ/mol

The relative free energy of the mutation Glu-->Ala = -4.732 kJ/mol

My main concern is with respect to the LJ transformation. Is my approach
correct in terms of modifying the side chain of glutamate to alanine? The
doubt arises because the relative free energy difference is negative where
as the experimental value is close to 20 kJ/mol. I am way under-predicting
the value and with a negative sign.

When I did these same calculations using AMBER 11, I am getting 6.945
kcal/mol which is still less but it does not have a negative sign.
The other thing I observed is the coulomb and the vdw dV/dl vs lambda
curves for OPLS-AA (gromacs) and AMBER 99SB (AMBER11) have a very similar
trend only shifted in the values of dV/dl in the y-axis.

I can attach the graphs and include more details, if needed.
Let me know.


Thanks for your time,

Regards
Sai



More information about the gromacs.org_gmx-users mailing list