this paper, we describe the construction of parametric surface skins using real spherical polar basis functions. As the use of such functions for protein shape representation is novel, a brief summary of their properties is also provided. We then give a description of the algebraic manipulations necessary to develop an efficient search for docking orienta- tions by incrementally rotating and translating the parametric representations. We also show that this spherical polar approach provides a natural way to model macromolecular electrostatic complementarity

otease-protein inhibitor and four antibody-antigen complexes. A large number of putative docked complexes have already been generated for the test systems using our rigid-body docking program, FTDOCK. They include geometries that closely resemble the crystal complex, and therefore act as a test for the renement procedure. In the protease-inhibitors, geometries that resemble the crystal complex are ranked in the top four solutions for four out of ve systems when solvation is included in the energy function, against a background of between 26 and 364 complexes in the data set. The results for the antibody-antigen complexes are not as encouraging, with only two of the four systems showing discrimination. It would appear that these results reect the somewhat different binding mechanism dominant in the two types of protein-protein complex. Binding in the protease-inhibitors appears to be "lock and key" in nature. The xed backbone and mobile side-chain representation provide a good model for

The ICM-DISCO (Docking and Interface Side-Chain Optimization) protein--proteindocking method is a direct stochastic global energy optimization from multiple starting positions of the ligand. The first step is performed by docking of a rigid all-atom ligand molecule to a set of soft receptor potentials precalculated on a 0.5 grid from realistic solvent-corrected force-field energies. This step finds the correct solution as the lowest energy conformation in almost 100% of the cases in which interfaces do not change on binding. The second step is needed to deal with the induced changes and includes the global optimization of the interface side-chains of up to 400 best solutions. The CAPRI predictions were performed fully automatically with this method. Available experimental information was included as a filtering step to favor expected docking surfaces. In three of the seven proposed targets, the ICM-DISCO method found a good solution (>50% of correct contacts) within the five submitted models. The procedure is global and fully automated. We demonstrate that the algorithm handles the induced changes of surface side-chains but is less successful if the backbone undergoes largescale rearrangements. Proteins 2003;52:113--117.

Motivation: Software systems predicting automatically whether and how two proteins may interact are highly desirable, both, for understanding biological processes and for rational design of new proteins. As a part of a future complete solution to this problem a bundle of programs is presented designed (1) to estimate initial docking positions for a given pair of docking candidates, (2) to adjust them, and (3) to filter them, thus preparing more detailed computations of free energies. Results: The system is evaluated on a test set of 51 cocrystallized complexes aiming at redocking the subunits. It works completely automatically and the evaluation is performed using one single set of parameters for all complexes in the test set. The number of solutions is fixed to 50 positions with a median CPU--time of 26min. For 30 complexes these contain a near--correct solution with RMSD 5.0 A, which is ranked first in five cases. For all complexes the best solution is scored on rank 16 as the worst case, and has a median RMSD of 4.3 A. Al- To whom correspondence should be addressed. Now with Behring Diagnostics GmbH, Instrumentation, Am Kronberger Hang 3, D--65824 Schwalbach/Ts., Germany. ternatively to this initial estimation of docking positions a global sampling of rotations was tested. Whereas this yields top--ranked solutions with RMSD 3.0 A for all 51 complexes the median CPU--time increases to 11h. This shows, that this blind sampling is not feasible for most applications. Availability: The system and its components are available on request from the authors. Contact: E-mail addresses: friedric@techfak.unibielefeld or posch@techfak.uni-bielefeld.de.

this paper we address two computational problems: prediction of the binding energy of both phosphorylated and unphosphorylated peptides given the geometry of the complex and prediction of the association geometry of a fully flexible peptide to its receptor. Theoretical binding energy evaluation is a complex and, as yet, unsolved problem [24-26]. Some methods use sophisticated and computationally costly approaches, such as molecular dynamics (MD) [27], free energy perturbation (FEP) and thermodynamic integration (TI) [28], protein dipoles Langevin dipoles model (PDLD) and linear-response approximation (LRA) [29]. Other methods use simplified binding functions that attempt to capture the major contributions to the 3 binding energy [24,30,31], such as polar and apolar contributions [32], entropy loss of the ligand [33,34], and surface complementarity [35]. In general, fast and simple evaluations of the binding energy are essential for a large scale screening of compound databanks [36]. The evaluation scheme used in this work was similar to the one used by Novotny and coworkers [33,34,37] with the exception of two modifications. First, we use a rigorous boundary-element evaluation of the electrostatic components of the interaction. Second, we took advantage of a rigorous sidechainrefinement procedure using Optimal-bias Monte Carlo-Minimization (OBMCM, also known as biased probability Monte Carlo [38]), that improves the accuracy of the electrostatic contribution (see Results and Discussion section). Flexible docking of peptides to their protein receptors remains an unsolved problem [39-41]. Even a single peptide in isolation has too many degrees of freedom for a reliable prediction. To avoid this difficulty, ligands are usually considered to be rigid or are constructed in...

ABSTRACT Mapping the epitope of an antibody is of great interest, since it contributes much to our understanding of the mechanisms of molecular recognition and provides the basis for rational vaccine design. Here we present Mapitope, a computer algorithm for epitope mapping. The algorithm input is a set of affinity isolated peptides obtained by screening phage display peptidelibraries with the antibody of interest. The output is usually 1–3 epitope candidates on the surface of the atomic structure of the antigen. We have systematically tested the performance of Mapitope by assessing the effect of the algorithm parameters on the final prediction. Thus, we have examined the effect of the statistical threshold (ST) parameter, relating to the frequency distribution and enrichment of amino acid pairs from the isolated peptides and the D (distance) and E (exposure) parameters which relate to the physical parameters of the antigen. Two model systems were analyzed in which the antibody of interest had previously been co-crystallized with the antigen and thus the epitope is a given. The Mapitope algorithm successfully predicted the epitopes in both models. Accordingly, we formulated a stepwise paradigm for the prediction of discontinuous conformational epitopes using peptides obtained from screening phage display libraries. We applied this paradigm to successfully predict the epitope of the Trastuzumab antibody on the surface of the Her-2/neu receptor in a third model system. Proteins 2007; 68:294–304. VC 2007 Wiley-Liss, Inc. Key words: antibody; combinatorial phage display peptide library; computational algorithm; epitope mapping; vaccine

Ligand-protein interactions are central and ubiquitous phenomena in a variety of biological processes from enzymatic catalysis to signal transduction. A theoretical understanding of

Introduction Identification of strain in protein three-dimensional structures has many important implications for both experimental structure determination and theoretical modeling and design. The term `steric strain' usually describes unfavorable or disallowed conformations or structural abnormalities of the amino acid residues that are detected by analysis of f/y maps, van der Waals clashes, etc. High-resolution X-ray crystal structures from the PDB [1] were analyzed [2,3] and it was shown that some parts of the polypeptide chain manifest higher strain, as a result of packing or functional requirements. An alternative and the most likely source of strain is the error in the coordinates of a structure, ranging from misplaced sidechains and flipped peptide groups to wrong chain tracing [4--6]. Similarly, the dynamic nature of a biopolymer in solution and the ambiguities in peak assignments may result in errors in the structures solved by NMR. Exam

this document were produced directly from this program

We describe a novel computer system directed to evaluate protein complex formation in a liquid environment. The relevant feature of the system is a potential function expressing the main thermodynamic and kinetic factors leading to protein interaction in solution. The protein interaction model expresses the interaction energy as basically composed of three forces: electrostatic (hydrogen bond), van der Waals, and hydrophobic. The latter is defined in function of the forces that the solvent molecules exert on the surface of the complex, and the van der Waals forces between the monomers and the solvent. The interaction model implemented in the system has proven a high discrimination ability between different protein dockings, scoring high those close to the observed crystal structures. These results have led to the establishment of the basic principles underlying protein interaction, which constitutes the main way of expression of the biological function of these macromolecules. 1