Recognition of binding patterns common to a set of protein structures

We present a novel computational method, MultiBind, for recognition of binding patterns common to a set of protein structures. It is the first method which performs a multiple alignment between protein binding sites in the absence of overall sequence, fold or binding partner similarity. MultiBind recognizes common spatial arrangements of physico-chemical properties in the binding sites. These should be important for recognition of function, prediction of binding and drug design. We discuss the theoretical aspects of the computational problem of multiple structure alignment. This problem involves solving a 3D k-partite matching problem, which we show to be NP-Hard. The MultiBind method, applies an efficient Geometric Hashing technique to detect a potential set of multiple alignments of the given binding sites. To overcome the exponential number of possible multiple combinations it applies a very efficient filtering procedure which is heavily based on the selected scoring function. Our method guarantees detection of an approximate solution in terms of pattern proximity as well as cardinality of multiple alignment. We show applications of MultiBind to several biological targets. The method recognizes patterns which are responsible for binding small molecules such as estradiol, ATP/ANP and transition state analogues. The presented computational results agree with the available biological ones

See More