SURF'S UP! is a web server for analysis of
functional relationships in protein families, as inferred from
protein surface maps comparison. It assigns a numerical score
to the similarity between patterns of physicochemical (charge,
hydrophobicity) features distribution on surfaces of pairs or
groups of proteins. This allows recognizing groups of proteins
within groups that have similar surfaces, hence presumably
By concentrating on coarse surface features, SURF'S_UP can work with models obtained from comparative modeling. Other valuable characteristic of our method is the lack of initial assumptions about the functional features that are to be compared.
The server takes as an input a set of superimposed protein coordinates and the user specifies features of protein surface that are to be compared. SURF'S UP! returns distance matrix and (optionally) unrooted tree as well as pictures of protein surfaces (in a tiff format). Each structure or structure model must be in the pdb format and should start eihter with the 'MODEL ' or 'COMPND ?' header (the second type of header is automatically generated by SwissPDBViewer program http://www.expasy.ch/spdbv/). The end of protein or protein model should be defined either by 'TER' or 'SPDBVe' line . An example file can be found here. Please notice that the input data are not supposed to be NMR models, but rather homology models or X-ray files.
The structures can be superimposed by the user (preferred
option), by a structure alignment program (such as for instance
available within the SwissPDBViewer program). In such case there would be
not made any rotation of protein structures.
When the uploaded structures are not superimposed, the spherical maps of proteins are "superimposed" while comparison procedure. For each pair of compared proteins the first map is chosen as a reference one. The second map is rorated around equator and meridian. The similarity between maps in each possible position is evaluated. The best score for each pair is taken as a final result.
"Features to compare" options
Amino acid residues can be divided into four classes, depending on their chemical features: hydrophobic (I, V, L, A, F, M, W, P, C), acidic (D, E), basic (H, K, R) and polar (N, Q, T, S, G, Y). The comparison of the proteins can be based
"Include side chain atoms" option
Only heavy (non-hydrogen) atoms of exposed residues are considered. The option of using only C-beta atoms of site-chains is recommended for low-resolution structures. Otherwise analysis with all the site-chain atoms should be performed.
Color images (in a tiff format) of protein surfaces represented by spheres are made using MOLSCRIPT program.
Choosing this option causes calculation of an unrooted tree (in a newick format).
"Send results by e-mail"
By filling in this textbox user mentions that he wants results to be sent by an email. The files that are to be sent to user are: distance matrix file, file with unrooted tree (if the option "generate tree" was chosen) and files with protein structures represented by spheres in pdb formats.
Can be downloaded here