PISA FAQ

  • PISA and Accessible and Buried Surface Area

    The 'View' 'Interface' option, using JMol, from the 'Interface Details' page, gives a excellent indication of the meaning of these terms. The non-interfacing atoms which are solvent accessible, are coloured light blue and dark blue for the two molecules respectively. Inaccessible atoms are coloured grey. These are more easily seen if you zoom into the display. (mouse wheel forward). It is apparent from this display that the solvent molecules would not fit into the holes where the grey atoms are. Interfacing atoms, (atoms which are part of interfacing residues), those atoms which are exposed to the other molecule and not the solvent, and may be involved in bonding across the interface, are coloured red and green respectively. Note: not all the atoms in an interfacing residue will be involving in interface formation or contact. Some may be exposed to the solvent and some may be inaccessible. This can be confirmed by hovering the mouse over a particular atom, when information about residue number and atom name is displayed. This is confirmed by inspection of the detailed table at the bottom of the 'Details' view.

    Accessible surface area calculations are based on finite element analysis (numerical analysis calculation) as a full exact analysis using integrals is "quite complex" even for something like water. A water probe of 1.4A in diameter is rolled over the surface of the protein and the sum of all sampled points in contact with this probe represents the surface area normalised by the precision of the element analysis.

    A second approximation is that the calculation uses enhanced radii to approximate the position of hydrogen atoms. This is clearly a significant aproximation as hydrogen atom positions are not isotropic around the parent atom, so increasing the radius of an atom to take account of the hydrogen create significant error.

    Finally - the calculation of the ASA of the interface is calculated as follows: (all numbers are A^2)

    If T is the total accessible surface area of the combined molecular components A:B where ":" is the interface between A and B, then you can see this total excludes the interface area.

    If TA and TB are the surface area of each molecular component A and B :-

    When A is in contact with B then the burried area of this contact of A MUST be the same as the area on B - since they are the same mutual contact.

    Therefore the missing area for A:B due to : is the same on A and B in isolation - so the measure of contact ":" missing in the A:B complex is counted twice - once in TA and once in TB

    Interface Area : = (TA + TB - T) / 2.0

    So if TA = 1000, TB = 2000, T = 2500

    interface area = ((1000 + 2000) - 2500) / 2.0 = 250

  • Table help information Most of the tables in PISA and Fold have headers which are also links, to pages of help on that column.
  • Calculations for delta G and delta H.

    The calculations used in PISA are described fully within the documentation and pages here... http://www.ebi.ac.uk/msd-srv/prot_int/picite.html and also A. SHRAKE AND J. A. RUPLEY 'Environment and Exposure to Solvent of Protein Atoms. J. Mol.Biol. (1997), 79, 351-371

    The calculations for delta G and delta H are very sensitive to approximations used though-out ;

    • There are no hydrogen atoms in most protein depositions which means an average enhanced (and spherical) approximation is used
    • Potential functions within proteins are not an exact science with electrostatic and non-bonding
    • interactions particularly problematic. In the former case this is very dependent on the dielectric constant which ranges from 80 in bulk solvent to 4 within the core of a protein and are highly variable based on local environment.
    • delta G and delta H are related based on the entropy of the system - which is normally estimated on the number of degrees of freedom - ie a guesstimate.
    • Water is a very difficult molecule to describe energetically, the fact it is a liquid at room temperature is a chemical anomoly; with a triple point at zero C. Most calculations are based on a dipole approximation though a quadrupole is slightly better. Since the association of protein chains is the absence of water on 2 protein surfaces, then the potential function is estimated using entropy, and this is a large guesstimate..

      Therefore the differences you see between programs are because of these estimations. The aim of PISA is to determine the likelyhood of assembly association and all calculations are optimised to produce a good estimate of likelyhood of correctly identify quaternary structure. I cannot comment on the FoldX program. It is very likely that these 2 programs result in different results.

    • Downloading as plain text. The current PISA program does not allow the 'Interfacing Residues' information to be downloaded as plain text.

      However, programs do exist to convert XML to CSV and other formats. Alternatively, if you have just a few entries to look at, you could copy and paste the information into a plain text editor.

    • Differences in Accessible Surface Area results from other programs The calculations of Accessible Surface Area (ASA) presented by PISA do not use any external packages.

      Some information is given in the documentation here: http://www.ebi.ac.uk/msd-srv/prot_int/picite.html and also A. SHRAKE AND J. A. RUPLEY 'Environment and Exposure to Solvent of Protein Atoms. J. Mol.Biol. (1997), 79, 351-371

      The method is one commonly used involving a probe (radius 1.4A) rolled around the residue atoms with enhanced radii.

      The calculations for delta G and delta H are very sensitive to approximations used though-out;

      1. There are no hydrogen atoms in most protein depositions which means an average enhanced (and spherical) approximation is used.
      2. Potential functions within proteins are not an exact science with electrostatic and non-bonding interactions particularly problematic. In the former case this is very dependent on the dielectric constant which ranges from 80 in bulk solvent to 4 within the core of a protein and are highly variable based on local environment.
      3. delta G and delta H are related based on the entropy of the system - which is normally estimated on the number of degrees of freedom - ie a guesstimate.
      4. Water is a very difficult molecule to describe energetically, the fact it is a liquid at room temperature is a chemical anomoly; with a triple point at zero C. Most calculations are based on a dipole approximation though a quadrupole is slightly better. Since the association of protein chains is the absence of water on 2 protein surfaces, then the potential function is estimated using entropy, and this is a large guesstimate.

      Therefore the differences you see between programs are because of these estimations. The aim of PISA is to determine the likelyhood of assembly association and all calculations are optimised to produce a good estimate of likelyhood of correctly identify quaternary structure. I cannot comment on the FoldX program. It is very likely that these 2 programs result in different results.

    • Hydrogen Bonding Criteria. PISA considers whether an H bond be present if the distance between the heavy atoms (donor and acceptor) is less than 3.89A. When the H atom is present the acceptor - H distance must be <= 4A and the angle A-H-D between 90 and 270.

      The relevant distance for a salt bridge is 4A. These distances are between the heavy atoms. The column headers also have help associated with them.

      The 'Details for each interface' (accessed by clicking on the number in the NN column) in PISA show that there is a wide variation in this distance. The number of H bonds and salt bridges is then used to assess the likely stability of the interface.

      Further detains are given here in citations here: http://www.ebi.ac.uk/msd-srv/prot_int/pistart.html

    • PISA Input Files. PISA is designed to analyse the contacts between molecules within a crystal. It will not predict the quaternary structure of a homology model which lacks crystal data within the file. For PISA to work it is necessary that the following remarks are present in the PDB file, additional to the atom data.
      HEADER    HYDROLASE (SERINE PROTEINASE)           22-JAN-85   5CHA      
      CRYST1   49.290   67.480   65.940  90.00 102.02  90.00 P 21          
      ORIGX1      1.000000  0.000000   .213065        0.00000                 
      ORIGX2      0.000000  1.000000  0.000000        0.00000                 
      ORIGX3      0.000000  0.000000  1.023106        0.00000                 
      SCALE1       .020288  0.000000   .004323        0.00000                 
      SCALE2      0.000000   .014819  0.000000        0.00000                 
      SCALE3      0.000000  0.000000   .015516        0.00000                 
      MTRIX1   1   .913870  -.006906   .405950       -9.94000    1            
      MTRIX2   1  -.006985  -.999918  -.012656       40.60000    1            
      MTRIX3   1   .406002   .010876  -.913807       47.60000    1            
      ATOM .........
      
    • Database Search Results. Example. Selecting Homotrimers. The result appears wrong in that 3QD6 is not returned in spite of having a homotrimer reported.

      PISA considers assemblies where the first PQS set has assemblies with differing multimeric states as 'ambiguous'. (which is a search term)

      In the Multimeric state drop down there is an option 'Ambiguous'. You can select more than one option using the Control key (on linux). So you can select Trimer + Ambiguous.

    • Browser The recommended browsers are Netscape and Chrome. The server should work with all browsers that support a primitive Java scripting. Support of frames is not needed.

      Please make sure that your browser is set to not using the cache. Go to browser's preferences, section "Cache", and set comparison to the network as "Every time" or "Automatically"

    • Backward and Forward Buttons Using these buttons in your browser is discouraged. PISA provides with all necessary in-screen navigation tools instead. Use of browser's navigation buttons, as well as the history bar, may result in confusion because the pages which you see, do not exist as html documents.
    • Interactive Work When PISA performs calculations on your requests, it displays a self-updating wait page showing the progress.

      If you do not see the progress: probably your browser reloads the wait page from the cache - please refer to section "BROWSER" above. If your browser does not allow switching the cache off, you will have to reload the wait page manually, e.g. by clicking on the "Reload" button of your browser.

    • Working in Parallel Sessions You may bookmark the wait page once it is displayed. This link would then take you either to results, if those are ready, or to the progress show. Even before your query finishes, you may submit another one by simply following the corresponding link in the wait page. Results expire in 4 hours after last access to them.