A Mathematical Framework for the Selection of an Optimal Set of Peptides for Epitope-Based Vaccines

Open Access
Publisher Website
Google Scholar
Abstract
Epitope-based vaccines (EVs) have a wide range of applications: from therapeutic to prophylactic approaches, from infectious diseases to cancer. The development of an EV is based on the knowledge of target-specific antigens from which immunogenic peptides, so-called epitopes, are derived. Such epitopes form the key components of the EV. Due to regulatory, economic, and practical concerns the number of epitopes that can be included in an EV is limited. Furthermore, as the major histocompatibility complex (MHC) binding these epitopes is highly polymorphic, every patient possesses a set of MHC class I and class II molecules of differing specificities. A peptide combination effective for one person can thus be completely ineffective for another. This renders the optimal selection of these epitopes an important and interesting optimization problem. In this work we present a mathematical framework based on integer linear programming (ILP) that allows the formulation of various flavors of the vaccine design problem and the efficient identification of optimal sets of epitopes. Out of a user-defined set of predicted or experimentally determined epitopes, the framework selects the set with the maximum likelihood of eliciting a broad and potent immune response. Our ILP approach allows an elegant and flexible formulation of numerous variants of the EV design problem. In order to demonstrate this, we show how common immunological requirements for a good EV (e.g., coverage of epitopes from each antigen, coverage of all MHC alleles in a set, or avoidance of epitopes with high mutation rates) can be translated into constraints or modifications of the objective function within the ILP framework. An implementation of the algorithm outperforms a simple greedy strategy as well as a previously suggested evolutionary algorithm and has runtimes on the order of seconds for typical problem sizes. Over the last decade the design of tailor-made vaccines for prophylactic applications (e.g., prevention of infection) and therapeutic applications (e.g., cancer therapy) has attracted significant interest. Epitope-based vaccines are good candidates for such tailor-made approaches. They trigger an immune response by confronting the immune system with immunogenic peptides derived from, e.g., viral- or cancer-specific proteins. These peptides bind to major histocompatibility complex (MHC) molecules in a specific manner. The resulting complex is crucial for immune system activation. However, there are many allelic variants of MHC molecules, meaning that different patients typically bind different repertoires of peptides. Nevertheless, due to economical and regulatory issues one cannot simply add all immunogenic peptides to such a peptide mix. Hence, it is crucial to identify the optimal set of peptides for a vaccine, given constraints such as MHC allele frequencies in the target population, peptide mutation rates, and maximum number of selected peptides. In this work we formalize this problem, and variants thereof, in a mathematical framework. The resulting optimization problem can be solved efficiently and yields a provably optimal peptide combination. We can show that the method performs better than existing solutions. Furthermore, the framework is highly flexible and can easily handle additional criteria.