Hybrid Quantum Mechanics / Molecular Mechanics (QM/MM) Approaches - - PowerPoint PPT Presentation

hybrid quantum mechanics molecular mechanics qm mm
SMART_READER_LITE
LIVE PREVIEW

Hybrid Quantum Mechanics / Molecular Mechanics (QM/MM) Approaches - - PowerPoint PPT Presentation

Hybrid Quantum Mechanics / Molecular Mechanics (QM/MM) Approaches -Treatment of the electrostatic QM/MM interface - Mauro Boero Institut de Physique et Chimie des Matriaux de Strasbourg University of Strasbourg - CNRS, F-67034 Strasbourg,


slide-1
SLIDE 1

1

Hybrid Quantum Mechanics / Molecular Mechanics (QM/MM) Approaches

  • Treatment of the electrostatic

QM/MM interface -

Mauro Boero

Institut de Physique et Chimie des Matériaux de Strasbourg University of Strasbourg - CNRS, F-67034 Strasbourg, France and @Institute of Materials and Systems for Sustainability, Nagoya University - Oshiyama Group, Nagoya Japan

slide-2
SLIDE 2

Treatment of the electrostatic in the QM/MM interface

  • Errors in the QM/MM Interface (OECP)
  • QM/MM interface: 3 level(s) coupling Hamiltonian
  • MM polarization

2

slide-3
SLIDE 3

From Gas-phase to complex environment

Molecules in the gas phase Solids and liquids

good properties reproduced using periodic boundary conditions (PBC)

  • Structure (radial distributions)
  • Dynamics (diffusion)

Complex disordered systems

No periodicity Partitioning of the system: QM/MM

No environment 3

slide-4
SLIDE 4

QM/MM Mixed Quantum-Classical

Partitioning the system: shopping list

  • 1. chemical active part treated by

QM methods

  • 2. large environment that is modeled

by a classical force field (MM)

  • 3. Interface between QM and classical

parts

QM/MM QM/Interface

  • A. Laio, J. VandeVondele, and U. Rothlisberger, J. Chem. Phys. 116, 6941 (2002);
  • A. Laio, J. VandeVondele, and U. Rothlisberger, J. Phys. Chem. B, 106, 7300, (2002);

4

slide-5
SLIDE 5

Partitioning of a QM system into 2 parts A and B: The non-linear (NL) correction

Approximations in QM/MM

QM/MM Mixed Quantum-Classical

where we use the “nuclear density”: and are unknown.

5

slide-6
SLIDE 6

QM/MM description of the subsystem B

Point charge representation of MM atoms at fix MM geometry

Because of the breakdown of the point charge representation: 2-, 3- and 4-body terms are needed:

QM/MM Mixed Quantum-Classical

Approximations in QM/MM

6

slide-7
SLIDE 7

QM/MM Mixed Quantum-Classical

QM description of A & MM description of B We collect all the approximations into the energy term

Approximations in QM/MM

7

slide-8
SLIDE 8

QM MM This term is small when

  • the electronic density is well localized within QM

( and are small), and

  • when we have a good force field for the MM part

 no bonds crossing the QM/MM boundary:  bonds crossing the QM/MM boundary:

crossing bond

QM/MM Mixed Quantum-Classical

We distinguish the two cases ( is small)

  • A. Laio, J. VandeVondele, and U. Rothlisberger, J. Chem. Phys. 116, 6941 (2002);
  • A. Laio, J. VandeVondele, and U. Rothlisberger, J. Phys. Chem. B, 106, 7300, (2002);

review in : M. Boero, Lect. Notes Phys. 795, pag. 81-98, Springer, Berlin Heidelberg 2010

Approximations in QM/MM

8

slide-9
SLIDE 9

We parameterize the potential using atom centered potentials, i.e. centered on the link atoms.

QM/MM Mixed Quantum-Classical

with Gaussian-type projectors, QM MM where Linking atom The parameters are determined through a fitting procedure. E CORR(RI,[])  drdr'i

*

(r)VI

OECP(r  RI,r'RI )i (r')

VI

OECP(r,r')

Ylm(r)pl(r)1

ml l

pl(r')Ylm

* (r')

pl(r)  rl exp(r2 /(2 2

2))

E CORR(RI,[])

O.A. von Lilienfeld, I. tavernelli, U. Rothlisberger, J. Chem. Phys. 122, 014113 (2005)

9

Approximations in QM/MM

slide-10
SLIDE 10

According to the Hohenberg-Kohn theorem

QM/MM Mixed Quantum-Classical

QM MM

“ the external potential is determined, within a trivial additive constant , by the electron density . This also determines the ground state wave function and all other electronic properties of the system”

In our QM/MM scheme we optimize the parameters

  • f the atom-centered potential in
  • rder to better reproduce the real quantum density in

the QM volume (obtained using a full QM description

  • f the total system). Thus we minimize the penalty

function:

O.A. von Lilienfeld, J. Chem. Phys. 122, 014113 (2005)

Approximations in QM/MM

10

Linking atom

slide-11
SLIDE 11

electronic density along z-axis Distance along the z-axis [Å]

z

QM/MM Mixed Quantum-Classical

11

slide-12
SLIDE 12

HOOC1-CH3 HOOC1-Dconv (1V) HOOC1-Dopt (DCACP) ESP (CH3/D) ESP (C1) Dipole [D]

1.67 2.87 1.41 0.00

[-0.30 + 3*0.1]

0.32

  • 0.03

0.74 0.28 0.54

Example: Acetic acid (Box size: 8 Å, gas phase, 80 Ry PW cutoff)

H O O H H H D O O H

QM/MM Mixed Quantum-Classical

O.A. von Lilienfeld, I. tavernelli, U. Rothlisberger, J. Chem. Phys. 122, 014113 (2005)

Approximations in QM/MM

12

slide-13
SLIDE 13

H tot[{i(r

i)},{RI}]  H DFT[{i(r i)};{RI}] H int[{i(r i)},{RI}] H MM [{RI}]

QM/MM Hamiltonian coupling additive scheme (just a reminder)

13

slide-14
SLIDE 14

QM/MM Hamiltonian coupling: Additive scheme

Scaling in a (not only) plane wave implementation:

H tot[{i(r

i)},{RI}]  H DFT[{i(r i)};{RI}] H int[{i(r i)},{RI}] H MM [{RI}]

H DFT[{i(r

i)};{RI}]: QM-part: Hartree and xc interaction O(Nel  NG  NG)

H int[{i(r

i)},{RI}]:

QM/MM interface O(N cl  NG)

H MM [{RI}]:

MM part: classical (ff) potential O(N cl  N cl) atoms classical

  • f

number the is functions set

  • basis
  • r

waves plane

  • f

number the is

cl G

N N

14

slide-15
SLIDE 15

H int (r),{qI,RI}

 

qI

I 1 Ncl

d3r (r) r  RI

Interaction Hamiltonian Potential acting on the QM electronic density Forces acting on the MM charged atoms:

E int[(r),{qI,RI}] (r)  qI r  RI

I 1 Ncl

V int(r)

 

int 3 3 int

) ( }] , { ), ( [

I I I I I I I

F R r R r r r d q R R q r E         

QM/MM Hamiltonian coupling: Electrostatics

15

slide-16
SLIDE 16

H int (r),{qI,RI}

 

qI

I 1 Ncl

d3r (r) r  RI

The nested sums (over the classical MM atoms and over the discretized QM volume) are in general computationally very expensive Ncl can be of the order of 100'000 - 500'000 Ngrid can be of the order of 200  200  200

QM/MM Hamiltonian coupling: Electrostatics

16

slide-17
SLIDE 17

QM/MM Hamiltonian coupling: The 3 regions scheme

Region 1: NN = subset of classical atoms inside the region Region 2: Classical-RESP charges interaction Region 3: Multipolar expansion on MM charges qI qJ

RESP(,RI )

RI  RJ

J QM

I NN

, r

1  RI  r 2

qI

I NN

[(r)] r  RI

 

r  RI

3 

, RI  r

2

qI

I 1 NN

d 3r

(r) r  RI , 0  RI  r

1

  • A. Laio, J. VandeVondele, and U. Rothlisberger, J. Chem. Phys. 116, 6941 (2002);
  • A. Laio, J. VandeVondele, and U. Rothlisberger, J. Phys. Chem. B, 106, 7300, (2002);

review in : M. Boero, Lect. Notes Phys. 795, pag. 81-98, Springer, Berlin Heidelberg 2010

17

slide-18
SLIDE 18

Generally we test:

However in all the known cases it is r1 ~ 10-12 a.u. r2 ~ 20-25 a.u.

Only NN < MM atoms in this shell

Divide the world in 3 domains: 1) Close to the QM region (r < r1) 2) Not too far, i.e. ESP region (r1 < r < r2) 3) Far MM world (r > r2)

r

1  r 2  

QM/MM Hamiltonian coupling: The 3 regions scheme

18

slide-19
SLIDE 19

R1: Direct coupling – the screened Coulomb potential

QM/MM Hamiltonian coupling R1: The direct coupling

E[(r),{RI}]  qI

ff

d3r

(r) r  RI

I NN

To avoid incompatibilities due to QM (electronic) vs. classical (point charges) description of the electrostatic, we introduce the screened Coulomb potential

E[(r),{RI}]  qI

ff

d3r

(r) vI

I NN

(r  RI )

where

vI (r)  RcI

n  rn

RcI

n1  rn1

is a “covalent” radius of atom I, and n is an integer (n=3)

RcI

n

19

slide-20
SLIDE 20

R2: D-RESP: Dynamical-Restrained ElectroStatic Potential derived charges

  • Define atomic point charges by fitting their value to the electrostatic

potential (ESP) due to the QM charge density seen by the close MM atoms

  • A restrain penalty function (RESP) is included, since unphysical

charge fluctuations have been observed in unrestrained ESP charges during dynamics. Namely, we minimize the norm

  qI

D

RI  RJ VJ

I QM

       

J NN

2

 wq qI

D  qI H

 

I QM

2

ESP RESTRAIN

qI

D = qI RESP

QM/MM Hamiltonian coupling R2: D-RESP region

20

slide-21
SLIDE 21

  qI

D

RI  RJ VJ

I QM

       

J NN

2

 wq qI

D qI H

 

I QM

2

is minimized on the fly during the dynamics. wq = weight parameter to reduce charge fluctuations:

VJ  d 3r(r)u

r  RJ

 

The potential VJ is the Coulomb potential generated on the MM atom J by the electronic density distribution

25 . 10 .  

q

w

where u(|r - rJ|) is a Coulomb potential modified at short range to avoid spurious over-polarization effects.

QM/MM Hamiltonian coupling R2: D-RESP region

21

slide-22
SLIDE 22

22

Hirshfeld charges ?

wq qI

D  qI H

 

I QM

In the restrain term qI

H are the so-called Hirshfeld charges. They are defined as

qI

H 

d3r r

 

I

at r  RI

 

K

at r  RK

 

K

 ZI

where at is the atomic (pseudo) valence charge density and

ZI  d3r at r  RI

 

is the bare valence charge of the I-th atom.

QM/MM Hamiltonian coupling R2: D-RESP region

  • F. L. Hirshfeld, Theoret. Chim. Acta,44, 129-138 (1977)
slide-23
SLIDE 23

  AI

JqI D  T J I

     

J

2

AI

J  1/RI  RJ

J  NN wqIJ J  QM    T J  VJ J  NN wqqJ

H

J  QM   

Note that a short-hand notation is used in which

AI

J = high index running on

= low index running on QM

 NN UQM

The Hirshfeld charges: the minimization of  is a least-square procedure where

23

QM/MM Hamiltonian coupling R2: D-RESP region

slide-24
SLIDE 24

The Hirshfeld charges least-square procedure:

 qI

D  0

 AK

JqK D  T J K

     

J

 AI

J  0

and the (analytical) solution is trivially

qI

D 

HIK

1tK K

where and HIK  AI

J AK J J

tK  AK

J T J J

QM/MM Hamiltonian coupling R2: D-RESP region

24

slide-25
SLIDE 25

VJ

RESP[(r)] 

qI

D[(r)]

RI  RJ

I QM

VJ

DFT[(r)] 

d3r(r)

u r  RJ

 

The D-RESP potential v RESP(r)  E RESP (r)  E RESP qI

D I QM

qI

D

(r) while the (extra) forces on the atoms are

FJ  rJ E RESP   E RESP RJ  E RESP qI

D I QM

qI

D

RJ

D-RESP potential - summary

can be computed in a much more efficient way than the exact DFT potential The corresponding potential on the electrons becomes

QM/MM Hamiltonian coupling R2: D-RESP region

25

slide-26
SLIDE 26

= center of the multipolar expansion (geometric center of the QM system)

R3: Multipolar (MP) expansion (actually quadrupolar) where

... ) )( ( 2 1 ) ( 1 | | ) (

5 , , , 3 , , 3

           

            

  

J I I z y x J I z y x J I

R r r R r R Q R r r R D R r C R r r r d

r

QM/MM Hamiltonian coupling R3: Multi-pole expansion region

26

slide-27
SLIDE 27

QM/MM Hamiltonian coupling Summary: R1 + R2 + R3

E QM /MM [(r),{RI}]  qI

ff

d 3r

(r) vI (r  RI )

I NN

 qI

D[(r)]qJ

RI  RJ

I QM

J MM (ESP)

C qJ RI  RJ

J MM(MP)

 D

x,y,z

qJ r RJ

3 J MM(MP)

(RI

 r )

 Q

,x,y,z

qJ r  RJ

5 J MM(MP)

(RI

  r )(RI   r )...

R1 R2 R3

27

slide-28
SLIDE 28

&QMMM COORDINATES INPUT TOPOLOGY ADD HYDROGEN AMBER ARRAYSIZES ... END ARRAYSIZES BOX TOLERANCE BOX WALLS CAPPING CAP HYDROGEN ELECTROSTATIC COUPLING [LONG RANGE] ESPWEIGHT EXCLUSION fGROMOS,LISTg FLEXIBLE WATER [ALL,BONDTYPE] FORCEMATCH ... END FORCEMATCH GROMOS HIRSHFELD [ON,OFF] MAXNN NOSPLIT RCUT NN RCUT MIX RCUT ESP RESTART TRAJECTORY SAMPLE INTERACTING [OFF,DCD] SPLIT TIMINGS UPDATE LIST VERBOSE WRITE LOCALTEMP [STEP fn ltg] &END

QM/MM Hamiltonian coupling Input file

28

slide-29
SLIDE 29

QM/MM Hamiltonian coupling Input file

ELECTROSTATIC COUPLING [LONG RANGE] Section: &QMMM

  • The electrostatic interaction of the quantum system with the classical system is explicitly kept

into account for all classical atoms at a distance R ≤ RCUT_NN from any quantum atom and for all the MM atoms at a distance of RCUT_NN < r ≤ RCUT_MIX and a charge larger than 0.1e (NN atoms).

  • MM-atoms with a charge smaller than 0.1e and a distance of RCUT_NN <r ≤ RCUT_MIX and

all MM-atoms with RCUT MIX < r ≤ RCUT ESP are coupled to the QM system by a ESP coupling Hamiltonian (EC atoms).

  • If the additional LONG RANGE keyword is specified, the interaction of the QM-system with

the rest of the classical atoms is explicitly kept into account via interacting with a multipole expansion for the QM-system up to quadrupolar order. A file named MULTIPOLE is produced.

  • If LONG RANGE is omitted the quantum system is coupled to the classical atoms not in the

NN-area and in the EC-area list via the force-field charges.

  • If the keyword ELECTROSTATIC COUPLING is omitted, all classical atoms are coupled to

the quantum system by the force-field charges (mechanical coupling). The files INTERACTING.pdb, TRAJECTORY_INTERACTING, MOVIE_INTERACTING, TRAJ_INT.dcd, and ESP (or some of them) are created. The list of NN and EC atoms is updated every 100 MD steps. This can be changed using the keyword UPDATE LIST. The default values for the cut-offs are RCU_ NN=RCUT_MIX=RCUT_ESP=10 a.u.. These values can be changed by the keywords RCUT_NN, RCUT_MIX, and RCUT_ESP with rnn ≤ rmix ≤ resp. 29

slide-30
SLIDE 30

The Redistributed Charge scheme (RC) is a scheme to improve the electrostatic at the linking atom.

  • Y. Zhang, H. Lin and D. G. Truhlar, J. Chem. Theory Comput. 3, 1378-1398 (2007)

QM/MM Hamiltonian coupling The Redistributed Charge scheme (RC)

It tunes the electrostatic balance between quantum and classical description at the interface.

30

slide-31
SLIDE 31

QM/MM Hamiltonian coupling The Redistributed Charge scheme (RC)

The RC scheme is used in conjunction with hydrogen capping approach Nomenclature for the MM atoms:

M1 M2x M2y M2z M3 M3 M3 Q1 Q2x Q2y Q2z Q3 Q3 Q3 31

slide-32
SLIDE 32

QM/MM Hamiltonian coupling The Redistributed Charge scheme (RC)

The RC scheme is used in conjunction with hydrogen capping approach The MM partial charge of atom M1 is redistributed evenly over 3 point charges q0=qM1/n, where n is the number of M1-M2 bonds. The capping hydrogen (or linking atom, HL) carries no charge. Location of the charges q0:

  • H. Lin and D. G. Truhlar, Theor. Chem. Acc. 117, 185-199

RM 1  Rq0 RM 1  RM 2  Cq0  0.5

32

slide-33
SLIDE 33

QM/MM Hamiltonian coupling The Redistributed Charge and Dipoles scheme (RCD)

The RC method introduces an error in the M1-M2 dipole. The dipole q0R (R=|RM1-RM2|) reduces by a factor of (1 – Cq0). For Cq0 = 0.5, the contribution reduces by 50%. In the RCD method, the values of redistributed charges q0 and of the charges

  • n M2 atoms (labeled k = 1, 2, …) are further modified such that these

contributions to the M1−M2 bond dipoles are preserved.

1 1

, 2 , 2 q q k M RCD k M q RCD

C C q q q C q q     

33

slide-34
SLIDE 34

QM/MM Hamiltonian coupling The Polarized-Boundary method (RCPB/RCDPB)

In the RC and RCD schemes, the QM subsystem is polarized by the MM system, but the MM subsystem is not polarized by the QM subsystem, resulting in an unbalanced treatment of the electrostatic interactions between the QM and MM subsystems. The polarized boundary RC (PBRC) and polarized-boundary RCD (PBRCD) schemes make improvements in this regard to the RC and RCD schemes, respectively, by allowing self-consistent mutual polarizations between the QM and MM subsystems near the boundary. In the PBRC and PBRCD schemes, the polarization of the MM subsystem due to the QM electrostatic potential is accomplished by adjusting the MM atomic partial charges in the QM/MM calculations according to the principles

  • f electronegativity equalization and charge conservations.

34

slide-35
SLIDE 35

QM/MM Hamiltonian coupling The Polarized-Boundary method (RCPB/RCDPB)

First, the MM atomic partial charges are assigned to the MM atoms, and the QM/MM calculation is performed. The electric field generated by the QM subsystem (nuclei and electronic wavefunctions) is computed and imposed on the MM, and a new set of MM atomic partial charges is determined according to the electronegativity equalization and charge conservation. The new set of partial charges replaces the old set of partial charges, and a new QM/MM calculation is performed with the updated partial charges (new external potential for the DFT calculation). convergence ?

STOP Convergence= until the variations in partial charges are smaller than preset thresholds, or until the number of iterations exceeds a preset value. no yes 35

slide-36
SLIDE 36

QM/MM Hamiltonian coupling The Polarized-Boundary method (RCPB/RCDPB)

Charge equalization method by Rappé and Goddard.

[A. K. Rappé and w. A. Goddard, W. A. (1991) J. Phys. Chem. 95, 3358-3363 (1991).]

EQ(Q

1,Q2,...,QNcl ) 

(EI 0  I

0QI  1

2 QI

2JII 0 ) I

 1 2 QI

I J

QJJIJ  QI d3r (r) r  RI

I

EI 0 is the unperturbed reference charge I

0 = E

Q      

I0

=1/2(IP - EA) is the first order response (electronegativity) term) Coulomb

  • self

(repulsive response

  • rder

second the is EA)

  • (IP

= Q E =

I0 2 2

         

II

J JIJ

0 Coulomb interaction between unit charges at I and J (at disrance R =| RI - RJ |)

Consider the classical energy expression:

36

slide-37
SLIDE 37

QM/MM Hamiltonian coupling The Polarized-Boundary method (RCPB/RCDPB)

The charge equalization method is based on the minimization of the energy

EQ(Q

1,Q2,...,QNcl )

i.e. under the constraints

) ,..., , (

2 1

        

I J J IJ J II I I N I

Q J Q J Q E Q Q Q

cl



  • 1. 1  2 K  Ncl
  • 2. Qtot 

QI

I1 Ncl

This leads to a linear system of equations CD  D where

D

1  Qtot, Di  i 0  1 0 for i  2

C1i  Qi, Cij  Jij  J1 j for i  2

37

slide-38
SLIDE 38

QM/MM Hamiltonian coupling The Polarized-Boundary method (RCPB/RCDPB)

Electronegativity equalization method by Mortier et. al.

[W. J. Mortier, S. K. Ghosh, and S. Shankar, J. Am. Chem. Soc. 108, 4315- 4320 (1986) .]

The existence of a unique chemical potential everywhere in the molecule establishes the electronegativity equalization principle that Mortier write as I (Q

1,Q2,...,QNcl )  (I 0  I )  2(I 0  I )QI 

QJ RIJ

JI

cl

N

      

2 1

with and are the neutral atom electronegativity and hardness, respectively, QI, and QJ are the charges on atoms I and J, and RIJ, is the internuclear

  • distance. The parameters and are the corrections to the neutral atom

electronegativity and hardness that arise as a consequence of bonding.

I I

I

 

I

38

slide-39
SLIDE 39

QM/MM Hamiltonian coupling The Polarized-Boundary method (RCPB/RCDPB)

and are obtained by calibrating through small-molecule calculations and are transferable and consistently usable for calculating charges in any molecule.

I I

39