· Delarue web servers
· Method's Flow chart
· Max. Likelihood Traj
  · Fit elastic constant
  · ProFit + Traj. job
  · NewPath Traj. job
  · All atoms reconstr.
  · Adenylate Kinase
  · Examples (Table)
· References
· Homes/Links
· Morphing (Linear)
· Job queue status

Generate the most probable trajectory between two states

Your email adress: (Recommended, for notification)

Job title: (Only alphanumerical characters)

PDB file of initial state (Maximum is 10000 atoms)
CA only for proteins, 3 atoms per residue for nucleic acids
Ex: 1ANF_ca.pdb

PDB file of final state (Maximum is 10000 atoms)
CA only for proteins. The two PDB files (initial and final) should have EXACTLY the same number of atoms.
Ex: 1OMP_ca.pdb

Length of the simulation (typically 50-200, in arbitrary units)

Number of snapshots along the path

Cutoff value for the ENM (default is 10 Angstrom)

Delta_E = E_start - E_final (in kcal/Mol)
See Figure below on the left for illustration

Elastic constant (left and right) in kcal/mol/Angstr**2
See Figure below on the right for illustration
A rule of thumb is that k*(cutoff)**2 ~ 10 kcal/Mol (roughly).
k_left and k_right
Add an extra scale factor (10x) for springs between successive C-alpha along the sequence ?

  • Additional analysis

    Overlap between the displacement vectors and the first normal modes along the trajectory (yes/no)
    Works fine up to 1600 atoms

    Transition time per residue: pseudo-'phi value' analysis (yes/no)
    See below for more details on pseudo-'phi values'.

  • Input data formats

    • The job title is just for your own identification, but note that it will show up in the public job queue (but your results will not be public).

    • The coordinate files should be in PDB format, with only a single structure (no multiple models). Only CA atoms are accepted as this stage for proteins; the limit is 10000 residues. A selection of 3-4 atoms should be used for nucleic acids.

      The two files MUST have the same number of (CA) atoms; here they will be structurally aligned internally.

    • The number of states is the number of snapshots output from the simulation. It should be of the order of 100.
    • The length of the simulation T should be such that the total action Stot in the log file does not change very much if T is increased (i.e. the user should check that the asymptotic regime has been reached).


    • In addition to the trajectory in Pymol format, the coordinates of the transition state are given, as well as the total action Stot and the energy of the transition state U_L (measured with respect to the energy of the left -initial- state).
    • Q1 and Q2 will also be found in the logfile, where Q1 is the number of contacts as in the initial state, and Q2 the number of contacts as in the final state, for each point along the trajectory.
      Here is an example (adenylate kinase), where the trajectory obtained with UMMS is also presented.
    • If needed, a Q1-specific - Q2-specific vs Q-common plot can also be built, upon request. This kind of plot reveals possible unfolding during the transition (see Okasaki et al., 2006).
    • The Elastic Energy during the transition could be calculated upon request. The attached plot demonstrates that MinActionPath generates a trajectory with less elastic energy than UMMS linearly interpolated one (see curve pink+green vs blue+red curve).
      This again concernes adenylate kinase with elastic spring constants of 0.1 kcal/Mol/Angstr^2 for both sides.
    • Finally, the question of the robustness of the trajectory with respect of the k1/k2 ratio, i.e. the elastic constants for both sides of the reaction, is of interest. Here is a Q1 vs Q2 plot for adenylate kinase and k2/k1=5, k1=k2 and k1/k2=2.

    Additional Outputs

    • An ASCII file ('something_overlap.out') will be generated. The first column corresponds to the frame number along the trajectory. The 106 following columns correspond, in ascending order, to the overlap between the set of difference vectors (with respect to the starting structure) and the N first normal modes (N=1,106, also calculated from the starting structure). The 6 first columns must always have negligible values compared to the 7th.
    • A PDB file ('something_pseudophi.pdb') will be generated. It corresponds to the PDB file of the end structure where the B-factor column is replaced with the pseudo-'phi value'. This allows for direct visualisation using either Pymol or VMD.
      To make a long story short, here we monitor the elastic energy of each residue (through its Calpha) along the trajectory. The 'time' at which this local elastic energy is maximal corresponds to its own 'transition time'. This is what is stored in the B-column. Comparing this value to the global transition time will tell you whether this residue has a early or late transition. Other definitions of residue-specific transition time(s) are described in Zhu and Hummer (2009), Biophys. J. 97:2456-63.

      Marc Delarue http://lorentz.dynstr.pasteur.fr