molecular dynamics of dna fragments on the grid

Molecular dynamics of DNAfragments on the Grid

Kirill Zinovjev

Latvian Institute of Organic SynthesisRiga Technical University

2009. g.

Molecular Dynamics

• Molecular system simulation on atomic level under given thermodynamic conditions (temperature, volume, pressure)

• Describe system evolution in time

• Capable for biological macromolecules (proteins, nucleic acids, membranes) in water solution

• Calculations are parallelizable (can be performed on the Grid)

Molecular dynamicsT=310° K, P=1 atm

System evolves in time

Theory

Thermodynamics (P,V,T)Newton equation of motion

Molecular mechanics (T = 0° K - absolute zero; P = 0 Pa - vacuo)

...

dihedral

angledistancemolecule

U

UUU

„E-box” biological function

• c-Myc-Max heterodimer binding → • TRRAP coactivator transporting to MBII domain →• Histone acetilation by HAT →

Gene activation and expression

RTU ETF Grid (Latvia)CYFRONET (Poland)

MethodsMolecular dynamics NMR NOESY

Varian Unity INova 600 MHz

Software

• NAMD – molecular dynamics calculation

• XPLOR – theoretical spectra calculation

• VMD – simulation system preparation, molecular dynamics trajectory visualization and analysis

Freeware!

NAMD, VMD – Open Source

Calculations

Input data(Structure, force field parameters, configuration file)

GRID(NAMD 2.6, MPI, 20-40 cores,

≈ 3000 CPU hours)

Temporary results, restart files

Every 3 hours

Storage≈ 6-8 GB each simulation

Final results(trajectory, log file, restart files)

Analysis (VMD, XPLOR)

Results

5’-CGCACGTGCG-3’

A4 incorrectly predicted NOE’s

E-box

0

100000

200000

300000

400000

0 50000 100000 150000 200000 250000 300000 350000

NOE (exp)

NO

E (

teor

)

Unique E-box features

• Distance between central nucleotides

5’-CGCAC(3.35 Å)GTGCG-3’• Unique sodium binding site• Increased hydration

Conclusions and problems

• Selected calculation approach satisfactory describe objects of interest and can be used to investigate the behavior of oligonucleotides in water solution

• The E-box sequence shows several sequence-selective features, that can be used to design substances with high E-box affinity.

• The calculations showed high parallelizability and were succesfully performed on the Grid

• Calculation errors are insufficiently described

• The simulation length is too short (10 ns)

• The CPU utilization dramatically falls, when processes are distributed between too much physical machines

• The calculation time strongly depends on data transfer speed between cores

Future

Molecular docking for medicinal chemistry

QM/MM simulations

Simulations of proteins, membranes and their complexes

Acknowledgements

BalticGrid project (www.balticgrid.org), especially

Janis Kulinsh and Lauris Cikovskis (RTU ETF)

Thank you!!!Thank you!!!