T-cell epitope prediction and immune complex simulation using molecular dynamics: state of the art and persisting challenges

Flower et al. Immunome Research 2010, 6(Suppl 2):S4 http://www.immunome-research.com/content/6/S2/S4 IMMUNOME RESEARCH REVIEW Open Access T-cell epitope prediction and immune complex simulation using molecular dynamics: state of the art and persisting challenges Darren R Flower1*, Kanchan Phadwal2, Isabel K Macdonald3, Peter V Coveney4, Matthew N Davies5†, Shunzhou Wan4 Abstract Atomistic Molecular Dynamics provides powerful and flexible tools for the prediction and analysis of molecular and macromolecular systems. Specifically, it provides a means by which we can measure theoretically that which cannot be measured experimentally: the dynamic time-evolution of complex systems comprising atoms and molecules. It is particularly suitable for the simulation and analysis of the otherwise inaccessible details of MHC-peptide interaction and, on a larger scale, the simulation of the immune synapse. Progress has been relatively tentative yet the emergence of truly high-performance computing and the development of coarse-grained simulation now offers us the hope of accurately predicting thermodynamic parameters and of simulating not merely a handful of proteins but larger, longer simulations comprising thousands of protein molecules and the cellular scale structures they form. We exemplify this within the context of immunoinformatics. Review As Donald Knuth, the famous polymath of modern-day computer science, wrote in the forward to the book “A = B”: Science is what we understand well enough to explain to a computer; Art is everything else we do. Whatever its verity, this statement has a clear and compelling appeal. The human mind, which has evolved within the macroscopic world to understand macroscopic phenomena and predict macroscopic behaviour, cannot completely grasp nor truly possesses a complete intuitive understanding of the microscopic world of atoms and molecules; a place which exists at the intersection of two worlds: the quantum world and the world of conventional, large-scale physics. In venturing to gain a scientific understanding of these elusive, recondite worlds far beyond the limits of our direct experience, we seek to do synthetic reductionism: dissecting phenomena associated with the nature and behaviour of the biological molecules comprising the systems we study, and then building mathematical models * Correspondence: † Contributed equally 1 Life and Health Sciences, Aston University, Aston Triangle, Birmingham B4 7ET, UK Full list of author information is available at the end of the article capable of predicting the more complex behaviour of the systems emerging from these components. It is only through accurate and robust prediction, that we can be sure that we have attained any degree of true understanding. The problem with bioscience in general and immunology in particular is that our knowledge of the basis of adaptive immunity is largely compiled from indirect sources. Such sources include in vitro experiments, which are performed in controlled yet often highly artificial conditions far removed from the biological context of the whole organism. To a harsh eye and a harsh mind, the interpretation of cellular function within the immune system is particularly indirect and inferential, being largely based on the use of flow cytometric detection of surface markers or the cytotoxic or proliferative behaviour of a bulk population of cells. All such experiments ultimately give insight only in a most circuitous manner. To pass to the practical, there has been much recent biomedical interest expressed in computational tools for the analysis of epitope-mediated immunogenicity [1-4]. The adaptive immune system saves us from the death and disability engendered by infectious disease. The adaptive immune response functions to destroy invading pathogens. © 2010 Flower et al; licensee BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Flower et al. Immunome Research 2010, 6(Suppl 2):S4 http://www.immunome-research.com/content/6/S2/S4 Effectively distinguishing foreign or non-self molecules from host or self molecules is vital. One half is the humoral immune response: antibodies, produced by B cells, bind to antigens on the surfaces of invading microbes. The cell-mediated immune response forms the other half of adaptive immunity; here activated T cells react against foreign antigen presented on the surface of other cells. Extensive, and continuing, work has been undertaken to develop novel epitope prediction methods, based on a variety of reliable and robust computational methods. The main focus has been the quantitative prediction of peptide-MHC interactions, the necessary precursor to the recognition of epitopes by T cell receptors, and the identification of continuous and discontinuous B-cell epitopes [5-8]. Such approaches seek to combine the best aspects of experimental and informatic science. Informatics, in the form of immunoinformatics, thus offers a considerable variety of tools and techniques for undertaking the rapid, robust, and accurate computational identification of epitopes. By using such in silico approaches, computer-based prediction methods can significantly increase the celerity of T-cell and B-cell epitope discovery, with a concomitant dividend for vaccine design and discovery. With an ever-increasing number of pathogen genomes now available, the mapping of B-cell and T-cell epitopes, both computational and experimental, is becoming a central issue in vaccine discovery [9-17]. However, using epitope mapping and epitope prediction makes understanding the structure or function of a particular pathogen gene essentially irrelevant. Nonetheless, gaining insight into function can add value to the exercise, allowing evolutionary rationales to be posited, for example. Mapping or prediction works solely with the physical structure of the protein, either in vivo, in vitro, or in silico, and insights into the role played by a protein in its organism of origin or within the context of host-pathogen interactions are simply not needed. By far the most successful prediction strategy has been the data-driven prediction of T-cell epitopes. In a pivotal retrospective analysis, Deavin et al. [18] compared several early direct T-cell epitope prediction methods, without finding a single method possessed of sufficient accuracy to be deemed useful. T-cell epitope prediction typically involves defining the peptide binding specificity of specific class I or class II MHC alleles and then predicting epitopes in silico. Today work focuses on predicting class I MHC-peptide binding affinity. At least for well-studied class I MHC alleles, such methods work well [19,20]. However, for prediction of all immune epitope data other than c (...truncated)