16
Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields The Effec(ve Fragment Molecular Orbital Method Jan H. Jensen and Casper Steinmann University of Copenhagen Dmitri Fedorov AIST, Japan JPC A 2010, 114, 8705 PLoS ONE 2012, 7:e41117 PLoS ONE 2012, 7:e44480 arxiv.org/abs/1212.6172 1

Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

Embed Size (px)

DESCRIPTION

Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

Citation preview

Page 1: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

Blurring  the  boundary  between  linear  scaling  QM,  QM/MM  and  polarizable  force  fields  

The  Effec(ve  Fragment  Molecular  Orbital  Method  

Jan  H.  Jensen  and  Casper  Steinmann  University  of  Copenhagen  

Dmitri  Fedorov  AIST,  Japan  

JPC  A  2010,  114,  8705  PLoS  ONE  2012,  7:e41117  PLoS  ONE  2012,  7:e44480  arxiv.org/abs/1212.6172  

1  

Page 2: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

The  Fragment  Molecular  Orbital  (FMO2)  method  (and  most  other  fragmentaEon  methods)  

2  

Page 3: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

The  Fragment  Molecular  Orbital  (FMO2)  method  (and  most  other  fragmentaEon  methods)  

Many-­‐body  PolarizaEon:

Monomer  SCF  in  the    Coulomb  field  of  all    other  monomers  

Iterated  to  self-­‐consistency    

3  

Page 4: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

The  Fragment  Molecular  Orbital  (FMO2)  method  (and  most  other  fragmentaEon  methods)  

Non-­‐Coulomb  effects:

Dimer  SCF  in  the    Coulomb  field  of  all    other  monomers  

Iterated  to  self-­‐consistency    

4  

Page 5: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

The  Fragment  Molecular  Orbital  (FMO2)  method  (and  most  other  fragmentaEon  methods)  

Coulomb  effects:    

Coulomb  energy  in  the    Coulomb  field  of  all    other  monomers  

5  

Page 6: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

The  EffecEve  Fragment  Molecular  Orbital  (EFMO)  method  (Using  ideas  from  the  EffecSve  Fragment  PotenSal  (EFP)  method)  

Monomer  SCF  in  the  gas  phase  

Extract  mulSpoles  and  dipole  polarizability  

6  

Page 7: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

The  EffecEve  Fragment  Molecular  Orbital  (EFMO)  method  (Using  ideas  from  the  EffecSve  Fragment  PotenSal  (EFP)  method)  

Many-­‐body  polarizaEon  

Computed  classically  using  induced  dipoles  for  enSre  system  

7  

Page 8: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

The  EffecEve  Fragment  Molecular  Orbital  (EFMO)  method  (Using  ideas  from  the  EffecSve  Fragment  PotenSal  (EFP)  method)  

Coulomb  and  Non-­‐Coulomb  effects  

dimer  SCF  in  the  gas  phase  

8  

Page 9: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

The  EffecEve  Fragment  Molecular  Orbital  (EFMO)  method  (Using  ideas  from  the  EffecSve  Fragment  PotenSal  (EFP)  method)  

Coulomb  effects  

Computed  using  staSc  mulSpoles  

9  

Page 10: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

MP2  (DFT  doesn’t  scale  well)  

+  0  

10  

Page 11: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

Covalent  FragmentaEon  (ElectrostaSc  screening  crucial)  

11  PLoS  ONE  2012,  7:e44480  

Page 12: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

Implemented  in  GAMESS  With  gradients  

Trp  cage  (20  residues)  2  residues/fragment  

                                                                                                   EFMO      FMO2  Error  in  energy                                                -­‐4.3                6.4    kcal/mol  

MP2/6-­‐31G(d)  gradient                  314              409    minutes  20  cores  (most  Sme  spent  in  MP2  dimers)  

12  

PLoS  ONE  2012,  7:e44480  

Page 13: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

QM/”MM”  

AcSve  

Frozen  

EEFMO = EA + EA /F + EF

EEFMO = EA0 + EAJ

0 − EA0 − EJ

0 − EIJPOL( )

J∈F

RI ,J ≤Rcut

∑ + EAJES

J∈F

RI ,J >Rcut

∑ + EtotPOL

h^p://arxiv.org/abs/1212.6172  

Page 14: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

14  

Proof-­‐of-­‐concept  arxiv.org/abs/1212.6172  

ONIOM:  MP2/cc-­‐pVDZ:EFMO-­‐RHF/6-­‐31G(d)  

16  Å  

ΔH ≠ = 18 vs 13 (exp)

4  days/path  80  CPUs  

Page 15: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

To  Do  MP2:RHF-­‐D  opSmizaSon  

EFMO/PCM  Flexible  EFP/Polarizable  “Force  Field”  

(EFMO  sSll  quite  slow  for  large  acSve  regions)  

EEFMO = EI0

I

N

∑ + EIJES + EIJ

XR /CT + EIJDisp( )

IJ

N

∑ + EtotPOL

EFMO  GUI:  FRAGIT.org  (Mikael  Ibsen)  

15  

Page 16: Blurring the boundary between linear scaling QM, QM/MM and polarizable force fields

Funding:    EU  (IRENE  collab  program)  

Thank  You!  

QuesEons  Now?  

QuesEons  Later?  

Leave  a  comment  on  

hZp://proteinsandwavefuncEons.blogspot.com/2013/01/new-­‐presentaEon-­‐blurring-­‐boundary.html  

16