In a protein we are working with, there is a 5 residues substructure of a beta-turn, in which we are really interested. First we would like to search and check whether similar structural elements are present in any other proteins, what is the best approach for this?
I would approach this by using a distance matrix-based method. First, translate your substructure into a distance matrix, i.e. calculate all pairwise C-alpha atoms distances. Second, represent each PDB structure as a distance matrix. Third, 'slide' your substructure-matrix over each PDB distance matrix. If there is a hit, this should be reflected by similar values in both matrices, i.e. the sum of differences should be close to zero.
The outcome and quality of matches of course depends on the chosen similarity threshold between two matrices to define structural similarity. First you may want to try it out with some known matches.
The same basic idea btw could also be found in the DALI algorithm by Lisa Holm.
Chris
I would just go for straight DALI search. The distance matrix approach requires coding and downloading the entire PDB.
http://mbs.cbrc.jp/papia-cgi/str_menu.pl
My guess is there are many more. Search google with strings like "protein 3d substructure fragment search"
The 2nd link is up. Here's ascreenshot: http://imageshack.us/photo/my-images/526/papia.png/
Log in to answer this question.
