Welcome to the protein structure quasi-rigid domain decomposition (PiSQRD) web server [1]. The server provides a user-friendly interface to apply the decomposition method introduced and described in ref. [2].

PiSQRD IN BRIEF

PiSQRD will subdivide single-chain or multimeric proteins into domains that behave approximately as rigid units in the course of protein structural fluctuations (occurring in thermal equilibrium). The best rigid-body decomposition is found using the lowest-energy collective modes of the system. By default the latter are calculated through the Beta-Gaussian elastic network model [3], or can be uploaded by the user.


[1] T. Aleksiev, R. Potestio, F.Pontiggia, S. Cozzini and C. Micheletti, PiSQRD: a web server for decomposing proteins into quasi-rigid dynamical domains. Bioinformatics, 2009 vol. 25, num. 20, pag. 2743-2744 Click here for on-line article.
[2] R. Potestio, F. Pontiggia and C. Micheletti, "Coarse-grained description of proteins' internal dynamics: an optimal strategy for decomposing proteins in rigid subunits", Biophysical Journal Volume 96 June 2009 4993–5002. Click here for on-line article.
[3] C. Micheletti, P. Carloni and A. Maritan, "Accurate and efficient description of protein vibrational dynamics: comparing molecular dynamics and Gaussian models", Proteins, 55, 635 (2004)


BASIC INPUT

In the default operation mode, the input required from the user is minimal:

  • the PDB code, or PDB coordinate file, of the protein of interest. If results are sought for a specific chain of a PDB entry, the chain identifier can be specified.
  • the desired fraction of internal dynamics that needs to be captured by the subdivision.

    • Notice that PiSQRD will not choose for you the number of returned domains. This is because the appropriate level of coarse-graining depends on the purpose of the subdivision. The correct compromise between coarse-graining and accuracy can be struck using as a guide the fraction of captured dynamics, f (provided in output).


    OUTPUT

    The PiSQRD engine will calculate all optimal subdivisions fro 2 to 20 domains.

    Upon successful completion, the user will be provided with the following output:

  • graph illustrating how the fraction of captured internal dynamics depends on the number of dynamical subdomains.
  • interactive graphical summary of the decomposition (provided through the jmol applet)
  • histogram representing the size (number of amino acids) taking part to the different domains
  • sequential representation of the domain assignment for each amino acid
  • dala (plain text) files containing the list of amino acids taking part to the different domains

    • By default, it is returned the subdivision corresponding to the smallest number of domains that is sufficient to capture the pre-assigned fraction of internal dynamics. At the bottom of the results page, users have the opportunity to select and visualize all other subdivisions, from 2 to 20 domains.

    ADVANCED INPUT

    In the advanced operation mode, the elastic network model calculation of the essential dynamical spaces can be overridden and users can upload their own data set. A single .tar file should be uploaded contaning, in two separate files, “covariance_eigenvalues.dat” and “covariance_eigenvectors.dat” the relevant dynamical information. Notice that if the provided files have names differing from above, they will not be processed! A brief diescription of the format of each of the two files follows.

    Eigenvalues file

  • name: covariance_eigenvalues.dat

  • format: the file must contain a list of the eigenvalues of the covariance matrix preceeded by an integer increasing index starting from 1, and must be ranked in decreasing order. Example:

    • 1    10.000
    2    9.000
    3    8.000
    4    7.000
    5    6.000
    6    5.000
    7    4.000
    8    3.000
    9    2.000
    10   1.000

    Eigenvectors file

  • name: covariance_eigenvectors.dat

  • The file must contain the eigenvectors' components (x, y, z) preceeded by the integer mode index. The consecutive ordering of the eigenvectors should follow the one of the eigenvalues. The eigenvectors of a protein of N amino acids, of which only the top L low-energy modes are considered, should be arranged as illustrated below (notice that different eigenvectors should not be separated by an empty line).

    • Both files must be included in a ".tar" file and uploaded to the server.

    For technical support please contact tyanko.alexiev@gmail.com