meshi – a new object oriented package for molecular simulations
DESCRIPTION
Meshi – a new object oriented package for molecular simulations Nir Kalisman, Ami Levi, Sharon Zafriri-lynn, Ram Janovski and Chen Keasar Departments of Computer Science and Life Sciences BGU. Challenges of computational structural biology. - PowerPoint PPT PresentationTRANSCRIPT
Meshi – a new object oriented package for molecular simulations
Nir Kalisman, Ami Levi, Sharon Zafriri-lynn, Ram Janovski and Chen Keasar
Departments of Computer Science and Life Sciences BGU
Challenges of computational structural biology
Computational determination of biomolecular structures.
Simulation of biomolecular dynamics.
Molecular design.
Computational tasks
Represent the system.molecular level?monomer (residue, nucleotide etc. ) level?
Atom level? All levels?
Measure structural properties.
Computational tasks (cont.)
• Evaluate structures
• Move from one structure to another.
Why a new package ?
Why Java ?
The modular, object oriented approach of Meshi.
Sample classes
Coordinates
CoordinatesX Y Z
ForcesFX FY FZ
Atom
type
name
coordinates
Distance
Z
d
Y
d
X
d
d
1atom1 atom2
d2
update
d
1
2
3
4
5
6
1 2 3 4 5
DistanceMatrix
update
Angle
ZYX
update
AngleEnergyElement
K
Atom type specific
constants
1.Evaluate
2.Update atom forces
2
Ke
AngleEnergy
1.Evaluate2.Update atom forces
A list of elements
eEangle
TotalEnergy
1.Update DistanceMatrix2.Evaluate
3.Update atom forces
term
termtotal eE
LBFGS(minimizer)
A list of coordinates
1.Update energy and forces
2.Change the coordinated
according to the forces.
3.If not yet in a minimum -
go to 1.
Current Meshi projects
Novel algorithm for non-bonded-list
generation.
MCMM-like stochastic optimization.
Continues knowledge-based potential for
pairs of torsion angles.
Cooperative hydrogen-bonds potential.
A refinement program for fold-recognition
models.
HO O
H
Novel algorithm for non-bonded-list
generation.
MCMM-like stochastic optimization.
Continues knowledge-based potential for
pairs of torsion angles.
Cooperative hydrogen-bonds potential.
A refinement program for fold-recognition
models.
HO O
H
1NXJ – T0156
native
Shotgun12.5Å (7.4)
Shotgun beautified11.1Å (6.5(Å
Package Tree
meshi
molecularElementsAtom Atomlist AtomPair AtomPairListResidue ResidueListProtein
geometryDistance DistanceMatrixAngleTorsion
energyAbstractEnergyTotalEnergyEnergyElementEnergyCreatorParametersParametersList.
optimizersSteepestDecentLBFGS
caCaProtein
backboneBackboneProtein
residuesUnitedAtomsUnitedAtomsResidueAla Cys Asp Glue….
bondBondEnergyBondEnergyElementBondParametersBondParametersListBondCreator
angleAngleEnergyAngleEnergyElementAngleParametersAngleParametersListAngleCreator
utilMeshiList
filtersFilter
PDB file format
PFRMAT TSTARGET T0201AUTHOR 3594-1086-1033METHOD Ab-initio with Mshi potential.ATOM 0 N MET 1 0.435 18.047 15.636 1.00 1.00ATOM 1 CA MET 1 -0.348 19.275 15.891 1.00 1.00ATOM 4 CB MET 1 -1.750 18.918 16.383 1.00 1.00ATOM 2 C MET 1 0.311 20.164 16.940 1.00 1.00ATOM 3 O MET 1 1.325 20.822 16.675 1.00 1.00ATOM 5 CG MET 1 -1.743 18.524 17.850 1.00 1.00ATOM 6 SD MET 1 -0.926 16.937 18.135 1.00 1.00ATOM 7 CE MET 1 -2.309 15.850 17.843 1.00 1.00ATOM 8 H ILE 2 -1.230 19.859 18.093 1.00 1.00ATOM 9 N ILE 2 -0.367 20.225 18.087 1.00 1.00ATOM 10 CA ILE 2 0.093 20.893 19.300 1.00 1.00ATOM 13 CB ILE 2 -0.961 21.864 19.867 1.00 1.00ATOM 11 C ILE 2 0.283 19.938 20.469 1.00 1.00ATOM 12 O ILE 2 -0.433 18.948 20.610 1.00 1.00ATOM 14 CG1 ILE 2 -2.157 21.058 20.375 1.00 1.00ATOM 15 CD1 ILE 2 -1.776 20.107 21.497 1.00 1.00ATOM 16 CG2 ILE 2 -1.417 22.904 18.851 1.00 1.00
parameters /users/faculty/cs/keasar/progs/meshi/current/meshi/parameters
bondEnergy weight 1LennardJones weight 1angleEnergy weight 1planeEnergy weight 1outOfPlaneEnergy weight 1000
minimize tolerance 0.01minimize maxSteps 500minimize reportEvery 10
A sample commands fileMinimizeProtein.prm