Complexity of Protein–Protein Interaction Networks, Complexes, and Pathways.pdf

451 31 Complexity of Protein–Protein Interaction Networks, Complexes, and Pathways Danail Bonchev 1. Introduction 1.1. The Network Approach to Proteomics The focus of proteomic re s e a rch in developing ex p e rimental techniques for pro t e i n i d e n t i fi c ation and interaction studies is shifting from individual proteins to their orga- n i z ation in reaction pat h way s , c o m p l exe s , and netwo rk s , i . e. , to the pro t e o m e — t h e l a rge-scale netwo rk comprising all pro t e i n – p rotein interactions in a cell, t i s s u e, o r o rganism. The number of complete proteomes in accessible dat abases exceeds 100 ( 1 ), thus making possible proteome-wide and acro s s - p roteomes analyses. Such a systemic ap p ro a ch offe rs a view of the biological machine as a wh o l e, revealing important new details of its wo rk. Th u s , one could rega rd aging and diseases as specific pat t e rns of p rotein netwo rk degra d ation and, vice ve rs a , evo l u t i o n a ry beneficial fa c t o rs as cre at- ing pat t e rns of larger proteome complex i t y. Medicines’ side effects could be analy ze d in terms of the ex t re m e ly high netwo rk connectiv i t y, thus orienting the search for new medicines towa rd protein complexe s , rather than individual compounds ( 2 ). Po t e n t i a l d rug and marker candidates could be identified proceeding from protein connectiv i t y and centrality pat t e rn s . 1.2. Quantifying Networks N e t wo rk analysis and ap p l i c ations are necessari ly re l ated to nu m b e rs that uniquely ch a ra c t e ri ze each netwo rk , making possible compari s o n , cl a s s i fi c at i o n , s t ru c t u re - a c t iv i t y / t oxicity re l at i o n s h i p s , and prediction. Rega rding netwo rks as graphs ( 3 ), one can use graph theory ( 4 , 5 ) to ge n e rate such nu m b e rs ,u s u a l ly termed t o p o l ogical indices ( 6 , 7 ), p roceeding from diffe rent graph inva riants. A va riety of topological indices have been p ropos