Homology modeling Molecular dynamics by xjl74245

VIEWS: 33 PAGES: 5

									                                                                                       Homology modeling
LECTURE 7                                      From step 2): the alignment of sequences is available
                                               E.g. homology modeling of trypsin IV (target) with 4CHA
                                               PDB file (template)
Contents

▪ Homology modeling
▪ Molecular dynamics
                       (Basic concepts)




                           Homology modeling                                           Homology modeling
                                               3) Backbone generation

                                               - Making a copy of the backbone coordinates of template
                                               protein.
                                               - Problems: PDB files are erroneous due to missing residues
                                               in many cases. Check of the template structure is necessary
                                               before and after the alignment.
                                               - Multiple template modeling can be used if no error-free
                                               templates are available (merging)

                                               4) Loop modeling

                                               - Gaps: in model sequence. Insertions: in template sequence
                                               - Gaps: a hole is created in template for gaps by deleting extra
                                               residues and closed later
                                               - Insertion: the template is cut and extra residues are inserted
                                         Homology modeling                                                Homology modeling
4) Loop modeling (cont.)
                                                                   - This works for high
- Both procedures require conformational change of backbone        sequence identitites.
- These modifications should be made in the loop regions as        At identity < 35 % the
in regular α-helix or β-sheet the conformational change of         rotamers of conserved
backbone is not permitted.                                         residues may differ in
- N.B. loops are highly flexible regions, often having high        up to 45 %.
B-factors. Conformation of a loop can easily vary in free state    - The core region of a protein
and in crystal structure (artifacts).                              often includes such conserved
                                                                   sequences and side-chain conformations can be predicted
5) Side-chain modeling                                             with high accuracy there.
                                                                   - Modeling of surface side-chains is more problematic (solvent).
- Coordinates corresponding to side-chain conformations at         - The side-chain conformations are dependent on the backbone
100 % conserved sequences can be copied very often from the        conformation, and, therefore knowledge-based construction
template into the modeled protein structure due to the             of side-chains often starts with alignment of backbones of
conservation of χ angles.                                          peptide fragments from known PDB structures.




                                         Homology modeling                                                Homology modeling
5) Side-chain modeling (cont.)
                                                                   Classification of homology models
- Other problems:
       - disulphide bridges (Cys-Cys): pre-defined parameters      1) Models based on incorrect alignments between target
         or constraints at model optimization.                           and template sequences. Low accuracy.
       - metal ion – side-chain contacts (Fe-S, Zn-S, etc.)        2) Models based on correct alignments are better, but their
       - modeling of hetero groups and co-factors                        accuracy can still be medium to low.
                                                                         Such models are very useful tools for the rational
6) Model optimization                                                    mutagenesis experiment design.
                                                                         They are of very limited use for ligand binding studies
Molecular dynamics, preferably with explicit solvent model.
                                                                         and rational drug design.
                                                                   3) Models of a high degree of sequence identity (> 70%)
7) Model validation                                                      with the target.
                                                                         Such models have proven useful during drug design
- Energy-based, using V of the MM force-field                            projects and allowed the taking of key decisions in
- Trial and error: reproduction of e.g. binding free energies of         compound optimization and chemical synthesis.
different ligands to the active site of the protein (docking)
                                              Molecular dynamics                                             Molecular dynamics
 Molecular dynamics (MD)                                           Why dynamics?

 - is a deterministic global search method                         Newton’s 2nd law formulates a basic principle of the
 - generates statistical amount of ensembles of                    dynamics of a body
    configurations, i.e. coordinates and velocities of all atoms   The rate of change of momentum of a body is proportional
    of the system                                                  to the resultant force acting on the body and is in the same
 - is a simulation method for calculation of microscopic           direction.
     and macroscopic properties of the systems
                                                                           r         d    r              r
 Some history                                                              Fi ( t ) = [mi v i ( t )] = miai ( t )
 1957 Alder és Wainwright, simple mechanical systems                                 dt
 1964 Rahman, Ar liquid
 1971 Rahman, water                                                    i: the body (atom); m: mass of the body; F: the force
 1977 McCammon és Karplus, protein in vacuum                           acting on i; a: acceleration; t: time
 1983 Gunsteren és Berendsen, protein in water




                                              Molecular dynamics                                             Molecular dynamics
How can we calculate the force?                                    Examples of calculation of forces

For a system with N atoms we know the potential energy             Non-bonding interactions
function (V) from molecular mechanics
           r r          r
        V( r1, r2 ,..., rN ) = Vbonding + Vnon −bonding

 For conservative forces (electrostatic, spring force, etc.)
 the force can be calculated as a gradient of the potential
       r        r                       ∂ ∂ ∂ 
       Fi = ∇V( ri )     wher e       ∇ , , 
                                        ∂x ∂y ∂z                  Bonding interaction
                                                 
  ∇: the nabla vector differential operator
 Integration                                            Molecular dynamics                                                            Molecular dynamics
                                                                                   An example: the Verlet algorithm
           r
r          Fi ( t )     r         d r       d2 r               r        dr                           r            r       r             r
ai ( t ) =            ; ai ( t ) = v i (t) = 2 ri (t)     ;    v i (t) = ri (t)                      r (t + ∆t) = r (t) + v(t)∆t + 2 a( t )∆t 2 + ...
                                                                                                                                     1
            mi                    dt        dt                          dt                           r            r       r             r
                                                                                                     r (t − ∆t) = r (t) − v(t)∆t + 2 a( t )∆t 2 + ...
                                                                                                                                     1


The motions of all particles                                                       By adding the two equations
are coupled. The many body                                                                            r              r      r              r
                                                                                                      r (t + ∆t) = 2r (t) − r ( t − ∆t ) + a( t )∆t 2
problem cannot be solved
analytically.                                                                      Velocities can be calculated from positions
                                                                                                      r         r              r
Finite difference methods                                                                             v( t ) = [r ( t + ∆t ) − r ( t − ∆t )]/ 2∆t
are used to generate MD
trajectories.
                                                                                    The Verlet algorithm requires 2 sets of atomic positions and
                                       2
r            r       dr           1 d
                                          r                                         one set of accelerations. Modest storage requirements.
r (t + ∆t) = r (t) + r (t)∆t + 2 2 r (t)∆t 2 + ...
                     dt             dt                                              Disadvantages: velocities are separately calculated and
r            r       r          r                        Taylor series expansion
r (t + ∆t) = r (t) + v(t)∆t + 2 a( t )∆t 2 + ...
                              1                            (index i is omitted)     it is not self-starting.




                                                        Molecular dynamics                                                            Molecular dynamics

 Recipe of an MD simulation (1 protein)
     INGREDIENTS
     - 1good-looking protein structure
          (preferably a PDB coordinate file)
     - Water
     - Some ions (for seasoning)
     DIRECTIONS
     Take a pretty simulation box which fits to the size of the
     protein. Fill up the box with water and add some ions
     until the charge becomes 0 (PME). Soak the protein in
     water until most H-bonds are formed. After equilibration
     put the system into an MD pot and start cooking. You
     can apply pressure and heat if necessary. The cooking
     time varies from ps’s to µs’s.
                                   Molecular dynamics                                              Molecular dynamics

Periodic boundary conditions                            Minimum image convention

- At the faces of the box                               Only the closest image of a molecule is considered
  water molecules have no                               during the calculation of short-range interactions.
  interaction partners, which
  is not the case in real
  solutions
- The box is surrounded by                              d > 2R c + Rprotein
   its images. Atoms at the
   faces interact with atoms                            d: edge of the box
   of the other image.                                  Rc: cut-off value of
- Molecules can leave at one                                 interactions
  side and enter at the opposite                        Rprotein: length of
                                                            protein along
  side to keep particle number
                                                            the edge
  constant.

								
To top