The formation of stable or transient protein-protein complexes plays a fundamental role in the majority of cellular processes. However, the structures of protein complexes are generally difficult to obtain via experimental methods, therefore computational approaches able to predict the reciprocal orientation of the interacting macromolecules and their interactions are becoming more and more useful. Protein-protein docking is one of the most used methods in this field, but still has some limitations, among them the impossibility to fully sample the conformational space of the complexes. Two strategies are mainly used to overcome this problem: treat the partners as rigid bodies or use restraints to limit the regions of the proteins to be sampled as interface. Here we present a new protein-protein docking strategy, ZADDOCK, that combines two docking methods, each using one of the two strategies, aimed at summing their strengths and limiting their drawbacks. As a first step, ZADDOCK employs the ZDOCK FFT-algorithm[1,2] to perform a rigid-body sampling of the conformational space of the complex. The resulting solutions are ranked with ZRANK,[3] and the best are refined by HADDOCK.[4,5] The latter is done in a two-step procedure consisting of a semi-flexible simulated annealing in torsion angle space, during which both side-chains and backbone at the interface are treated as flexible, followed by a final refinement in explicit solvent. To evaluate this combined approach, ZADDOCK was applied to a docking benchmark[6] composed of more than one hundred different cases. These sample different types of complexes (enzyme/inhibitors, antibody/antigen and others) and represent different levels of docking difficulty,[7] from “rigid” complexes for which none of the components undergoes significant conformational changes upon binding, to more challenging cases, for which the binding causes large conformational changes. Our results showed that the performance of this method is on average better or comparable to the performance of both ZDOCK+ZRANK and HADDOCK (in its “ab initio” mode) and does not depend on the type of complexes considered. In particular, very good results were obtained for those complexes that undergo minor conformational changes during the binding process. Moreover, the native contacts are better reproduced than was observed with ZDOCK. This indicates that ZADDOCK combines the strengths of ZDOCK and HADDOCK in its ability to almost fully sample the relative orientations of the two partners and to optimize the structure of a complex in a physically-based way.
Bordogna, A., Bonati, L., Bonvin, A. (2010). ZADDOCK: incorporating ab initio search into HADDOCK. Intervento presentato a: From Molecular Structure to System Biology - Convegno Nazionale della Divisione di Chimica dei Sistemi Biologici della Società Chimica Italiana, San Vito di Cadore.
ZADDOCK: incorporating ab initio search into HADDOCK
BORDOGNA, ANNALISA;BONATI, LAURA;
2010
Abstract
The formation of stable or transient protein-protein complexes plays a fundamental role in the majority of cellular processes. However, the structures of protein complexes are generally difficult to obtain via experimental methods, therefore computational approaches able to predict the reciprocal orientation of the interacting macromolecules and their interactions are becoming more and more useful. Protein-protein docking is one of the most used methods in this field, but still has some limitations, among them the impossibility to fully sample the conformational space of the complexes. Two strategies are mainly used to overcome this problem: treat the partners as rigid bodies or use restraints to limit the regions of the proteins to be sampled as interface. Here we present a new protein-protein docking strategy, ZADDOCK, that combines two docking methods, each using one of the two strategies, aimed at summing their strengths and limiting their drawbacks. As a first step, ZADDOCK employs the ZDOCK FFT-algorithm[1,2] to perform a rigid-body sampling of the conformational space of the complex. The resulting solutions are ranked with ZRANK,[3] and the best are refined by HADDOCK.[4,5] The latter is done in a two-step procedure consisting of a semi-flexible simulated annealing in torsion angle space, during which both side-chains and backbone at the interface are treated as flexible, followed by a final refinement in explicit solvent. To evaluate this combined approach, ZADDOCK was applied to a docking benchmark[6] composed of more than one hundred different cases. These sample different types of complexes (enzyme/inhibitors, antibody/antigen and others) and represent different levels of docking difficulty,[7] from “rigid” complexes for which none of the components undergoes significant conformational changes upon binding, to more challenging cases, for which the binding causes large conformational changes. Our results showed that the performance of this method is on average better or comparable to the performance of both ZDOCK+ZRANK and HADDOCK (in its “ab initio” mode) and does not depend on the type of complexes considered. In particular, very good results were obtained for those complexes that undergo minor conformational changes during the binding process. Moreover, the native contacts are better reproduced than was observed with ZDOCK. This indicates that ZADDOCK combines the strengths of ZDOCK and HADDOCK in its ability to almost fully sample the relative orientations of the two partners and to optimize the structure of a complex in a physically-based way.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
