Transcript presentation - Bulgarian Supercomputing Centre

Life science and nanotechnology software
applications
L. Litov, P. Petkov, G. Vayssilov
University of Sofia
Outlook






Physics basics
Quantum simulations
Molecular dynamics
Examples – nanotechnology
Examples – life science
Summary
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Physics basics
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Physics
x(t )
Newton equation
d
v(t )  x(t )
dt
d2
F  m 2 x(t )
dx
Schrödinger equation
Probability to find
P( x, t ) = Ψ* ( x, t )Ψ( x, t ) the
system in (r,t)
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Quantum simulations
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Quantum simulations
Schrödinger equation
Born- Oppenheimer approximation










 




(
R
,
R
,
R
...
R
,
r
,
r
,
r
,...
r
)


(
R
,
R
,
R
...
R
)

(
r
,
r
,
r
,..
r
)
1
2
3
N
1
2
3
n
nu
1
2
3
N
el
1
2
3
n
Schrödinger equation for electrons
   
   
 el el (r1 , r2 , r3 ,...rn , R)  Eel (r1 , r2 , r3 ,...rn , R)
me≪ m p
2 Z
I1



1
/
2





el
i
R
r
rij
i
I
,
i

j
I
ii
Schrödinger equation fore nuclei
  

  

H nu nu ( R1 , R2 , R3 ....RN )  Etot nu ( R1 , R2 , R3 ....RN )
Z
Z
12
I
J





E
(
R
)



nu
I
2
M
R
I I
I
,
J
IJ
L. Litov
E (R )
Life science and nanotechnology software applications
Sofia, 10 December 2010
Quantum simulations
 Hartree – Fock approximation - huge computational time
 Density Functional Theory (DFT) - Hohenberg-Kohn, Kohn- Sham
 Ground state properties of many electron system are determined by electron
density
 Interacting electrons in static external potential  non-interacting electrons in
effective potential
VASP, CP2K, CPMD, GAMESS, GAUSSIAN, Q-Chem
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Molecular Dynamics
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
From quantum to classic mechanics
For particle with mass m in equilibrium and temperature T, the mean value of the
momentum is
Heisenberg uncertainty principle
Follows that the width is
It is possible to omit the quantum effects if the variations sx are less than some critical width defining
the accuracy of our calculations..
Critical width
L. Litov
Моделиране на взаимодействия на
биологични молекули
Life science and nanotechnology software applications
Sofia, 10 December 2010
Quantum – Classic mechanics
• Electrons are treated by means of quantum mechanics
• Hydrogen and Deuterium atoms at temperature 300 K can not be
considered as a pure classical objects
• Heavier atoms can be treated as classical objects (temperatures 300
K). Additional correction are introduced in order to take into account
quantum effects.
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Solving the equations of motion
Point in the phase space of the system
Hamilton equations
- Liouville operator
H  K  V   pi2 / 2mi  V ( x)
where
i
In Cartesian coordinates
и
L. Litov
Do not commutate
Life science and nanotechnology software applications
Sofia, 10 December 2010
Solving the equations
Time propagator
One step propagator
Applying
on
We obtain
Velocity Verlet, Leap Frog algorithms
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Potential – force field
Every atom should be included
-chemical bonds with other atoms
-long distance interactions
V  Vs  Va  Vt  Vv  Ve  ...
i
Bond
strength
Sum
over all
bonds
L. Litov
i
Bond
angle
Sum
over all
angles
i
Torsi
on
i
i
Van der
Waals
interactio
ns
Life science and nanotechnology software applications
Coulomb
interaction
Sofia, 10 December 2010
Potential – force field
Parameterization of chemical bonds
vb r   12 kb r  r0 
2
va θ  12 ka θ  θ0 
2
vd φ   kd 1  cosnφ  φ0 
V x1 , x2 ,, xN   Vk x; pk 
k
- Empirical parameters (pk)
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Potential – force field
Parameterization of the other interactions
Vlj rij  
Cij1 2
rij1 2

Cij 6 
rij6
Vc rij  
qi q j
4 πε0 rij
V x1 , x2 ,, xN    vk x; pk 
k
- Empirical parameters (pk)
AMBER, CHARMM,GROMOS, GROMACS, LAMMPS, NAMD, VMD
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
New materials - nanothenology
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
nanothenology
 Two examples
 Interface between Pt8 cluster and Ce21O42 nanoparticle
 Reduction effect - generation of Ce3+ cations
 Effect on formation energy of oxygen vacancies
 VASP
 Hydrogen reverse spillover on zeolite-supported clusters
 Ab initio MD simulation of Rh4 and Ag4 clusters
 CP2K
Georgi N. Vayssilov, P. Petkov, H. Aleksandrov (Univ. of Sofia)
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Platinum cluster on ceria nanoparticle
 Pt/CeO2 - the key component of the automotive catalyst
 Model: cluster Pt8 on Ce21O42 nanoparticle
 Reduction of one Ce4+ to Ce3+ in the most stable structures
Pt
Ce3+
Eads = -5.03 eV
Ns = 4
O2-
1 Ce3+
Ce4+
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Clusters Ce21O42
 Plane-wave density-functional calculations

Model clusters Ce21O42

VASP code

PW91 gradient-corrected functional + U = 4 eV

Plane wave basis, cutoff of 415 eV

Spin-polarized calculations (where appropriate)

Unit cells: 202020 Å, allowing ~10 Å vacuum between
neighboring cluster images
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Platinum cluster on ceria nanoparticle
 Energy for formation of an O vacancy Ef is reduced in the presence of
platinum: ΔEf = 0.44 eV
-1/2 O2
1.67 eV
-1/2 O2
1.23 eV
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Ab initio MD simulations of supported clusters
 Support - Zeolite MOR - total 295 atoms per unit cell
 a = 18.256, b = 20.534, c = 15.084 A
 angles = 90.0
 Si/Al = 89/7 ≈ 13
 Initial M-H distances: ~250, ~275, ~470 … pm
Rh4
L. Litov
Ag4
Life science and nanotechnology software applications
Sofia, 10 December 2010
Ab initio MD simulation of Rh4 and Ag4 clusters
 Periodic ab initio MD simulations – CP2K
 DFT: PBE; BO MD and optimization
 PW basis, 200 eV cutoff for MD and 400 eV for geometry
optimization
 NVT ensemble
 MD run: time step 1 fs; 1 frame = 10 fs; time ~20 ps
 T = 300 K; CSVR thermostat
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
10ps MD run: Proton transfer to Rh4
Protons to be
transferred
L. Litov
Life science and nanotechnology software applications
23
Sofia, 10 December
2010
10ps MD run: Interaction with OH for Ag4
L. Litov
Life science and nanotechnology software applications
24
Sofia, 10 December
2010
Life science
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Understanding of human
interferon-gamma binding
Human Interferon Gamma
Active site
Res 18-26
Active site
Res 18-26
N-terminus
N-terminus
C-terminus C-terminus
122-143
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Interferon-gamma and its alpha receptor
PDB ID: 1fg9
Residues connected by H-bonds
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Gamma interferon binding
Task
 To find a possible way to inhibit the gamma-interferon activity
 Block the binding sites of the gamma-interferon
Find a ligand binding hIFN-g and blocking its activity
 Block the binding receptors (hIFNgRa) on the cell surface
With mutated hIFN-g peptides, lacking biological activity
With some other ligand
 Need to understand the mechanism of hIFN-g binding to its
cell receptors
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
hIFNg + hIFNgRa in water 26 ns
High performance computing, large
scale simulation and
drug design
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
INF-g C-terminus
D1 domain (125KTGRKRKR132)
D2 domain (137RGRR140)
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
INF-g C-terminus
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
hINF-g  hIFNgRa interaction simulations
GROMACS
High performance computing, large
scale simulation and
drug design
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
INF-g C-terminus
Heparin derived oligosaccharide
Biochem J. 2004 November 15; 384(Pt 1): 93–99.
NMR characterization of the interaction
between the C-terminal domain of interferon-γ
and heparin-derived oligosaccharides
Cécile Vanhaverbeke,*1 Jean-Pierre Simorre,* Rabia Sadir,†
Pierre Gans,*2 and Hugues Lortat-Jacob†
PDB ID: 1hpn
High performance computing, large
scale simulation and
drug design
dp8
dp4
dp2
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
hIFN-g and d8
High performance computing, large
scale simulation and
drug design
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Ribosome structure model
 The goal - construction and verification of a stable full atomistic computer
model of the whole ribosome, which enables realistic simulations of various
biochemical processes in the living cell.
 Stable ribosome subunits
 Construction of the whole ribosome including tRNA, mRNA, and a growing
peptide chain
 Determination of the structure of the ribosome in water
 Investigation of the influence of the type of the cations on the stability of the
whole ribosome (role of Na and Mg ions)
 Non trivial challenging task requiring a Petascale (~3.106 atoms) computing
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Ribosome big subunit





CHARMM27 Force Field
Explicit solvent MD simulation
Crystallographic ions included
NAMD 2.6

Time step 2.5 fs

NVT
Unstable structure even in about
0.5 ns long MD simulation
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Ribosome big subunit with Na+ counter ions

Na+ ions added to crystallographic
structure compensating the charge
of phosphates
Na+
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Trajectories RMSD
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
COX inhibitors




Investigation of the system – enzyme- inhibitor
Cyclooxygenase (COX 1 and COX I2) – responsible for many cell
processes
 Regulation and production of hormones
 Regulation of the Ca transfer
 Thrombosis aggregation
 Regulation of inflammatory processes ..etc.
COX1 and COX2 bind with arachidonic acid – produces
prostaglandines
COX1 and COX2 are targets for all nonsteroidal anti- inflammatory
drugs like aspirin, paracetamol etc.
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Docking
 Docking is based on:
 Search of the most
suitable ligand
orientation with respect
to the receptor centre
 Define the binding
affinity – using different
scoring functions
 Calculation of the
binding energy
 DOCK 6.4
 Selectivity test
 512 ligands
 There is no binding
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Arachidonic acid
Arachidonic acid binding is reproduced correctly (crystallographic structure)
RMSD = 1,820 Å;
ECryst = 55,45 kcal/mol;
ΔE = 4,2 kcal/mol;
L. Litov
EDock = 51,25 kcal/mo
Life science and nanotechnology software applications
Sofia, 10 December 2010
COX1 and COX2 inhibitors
 Investigation of inhibitor ligands
 Crystallographic orientation of
some inhibitors are reproduced
well
 Ibuprofen, Fluribiprofen etc
 Binding of specific COX2
inhibitors is under investigation
 Diclofenak, Celecoxib
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Conclusions
 Computer simulations:
 are extremely useful in the design of new materials
 can play significant role in the understanding of biological processes at atomic
and molecular level
 significantly reduce time and cost of development of new drugs (in-silco drug
design)
 Quantum calculations are extremely time consuming – require new algorithms and
more powerful (super)computers.
 Simulations of large (milions of atoms) systems require supercomputing at
Petascale level
 Problem with scaling – require new algorithms in order to reduce procesor
communications
 Reach variety of software is installed on Bulgarian IBM BG/P supercomputer
 We are welcome to run your jobs at BG supercomputer
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Spare slides
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Algorithms for solving the equation of motion
Error at every step
Accumulated error
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Systematic errors
 Force field (Gromacs, NAMD)
 Water model – different for NAMD and GROMACS
 Box size
 Periodic boundary conditions
 Need special investigations
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Platinum cluster on ceria nanoparticle
 DOS plots of bare CeO2 and Pt8/CeO2:
Pt states appear in the gap between O 2p and Ce 4f “bands”
EFermi
CeO2
- dashed lines
Pt8/CeO2 - solid lines
L. Litov
Pt states
Life science and nanotechnology software applications 51
51
Sofia, 10 December
2010
Energies for proton transfer to Rh4
Rh4
H
ERS, kJ/mol
Rin(M-H), pm
1
-272
250
2
-103
275
3
20
475
Proton transfer
from distant OH
group is
disfavored
L. Litov
Energies obtained
after optimization of
the structures with
different number of
transferred H
Proton transfer from
near OH group to the
metal cluster is
favored
Life science and nanotechnology software applications
52
Sofia, 10 December
2010
Systematic errors - models
NAMD
L. Litov
GROMACS
Life science and nanotechnology software applications
Sofia, 10 December 2010
Systematic errors – box size
High performance computing, large
scale simulation and
drug design
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Systematic errors - PBC
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
Systematic errors - PBC
High performance computing, large
scale simulation and
drug design
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
High performance computing, large
scale simulation and
drug design
L.L. Litov
Second
Life science and nanotechnology software applications
workshop
Sofia, 10 December 2010
INF-g C-terminus – dp8 interaction
Configuration 1
Configuration 2
T = 0 ps
High performance computing, large
scale simulation and
drug design
T = 2000 ps
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010
INF-g C-terminus – dp8 interaction
High performance computing, large
scale simulation and
drug design
L. Litov
Life science and nanotechnology software applications
Sofia, 10 December 2010