We then recalculated the ranking of each paratope only against paratopes in the same cluster, hence with similar PCs (Structurally Similar Antibody Rank). power both in terms of identifying the cognate antigen target for a given antibody and the antibody target for a given antigen, exceeding the results of other available tools. Keywords:antigen, antibody, SHR1653 B cell epitope, prediction, paratope, antibody specific epitope prediction == Introduction == B-cells form an essential part of the adaptive immune system, as they are capable of providing long-term protection against pathogens and harmful molecules. Their extremely specific B-cell receptors, named immunoglobulins or antibodies, are key components in this process. Antibodies recognize their molecular targets, termed antigens, via interactions between their binding site (paratope) SHR1653 and a specific region of the antigen (epitope). Most B-cell epitopes are discontinuous in sequence, meaning that they are composed of residues that might be far apart in sequence and are brought together in spatial proximity by the protein folding (1). Data describing such conformational epitopes are mainly obtained from experimentally resolved 3D structures of antibodies co-crystallized with their target antigen, which allows a very precise identification of the epitope residues. Identification of B-cell epitopes is of high importance for many medical, immunological and biological applications including disease control, diagnostics, and vaccine development (2). Several experimental methods for epitope identification are available including protein crystallography, ELISA and peptide-chip, but in general they are expensive, time consuming, low-throughput, or have low accuracy. Several computational methods have been developed to assist or substitute the experimental approaches, including BepiPred, DiscoTope, CBtope, and ABCpred (36). In absence of information on the cognate antibody, B-cell epitope prediction tools can broadly be categorized in two groups: sequence- and structure-based methods. As the names suggest, sequence-based methods predict the B-cell epitopes from the protein sequence of the antigen alone, whereas structure-based methods take into account also their 3D structure. Many benchmark studies have, as expected, demonstrated that structure-based methods display superior performance compared to sequence-based methods (7). However, even the best current structure-based methods for B-cell epitope prediction have limited predictive power (4). One important reason for this is that in many cases the problem is definitely ill-posed. If we arranged the task in very broad terms, we goal at predicting whether a given surface patch of an antigen is definitely a potential epitope, i.e., if one or more of the many billion antibodies potentially present in a host can target this region. Formulated like this, most surface patches of an antigen are potential epitopes. This has a serious impact on the way we define an appropriate dataset for teaching and evaluating a B-cell epitope prediction method, and on the limited predictive power of B-cell epitope prediction method (4). Moreover, in many applications, it is often more important to understand which region of an antigen can be targeted by a specific antibody, or by a group of antibodies, e.g., a library or an antibody repertoire acquired via Rep-Seq (8). This important observation has led to alternative and more well-defined approaches becoming proposed to address B-cell epitope prediction, where one seeks to forecast the cognate target of a given antibody (9,10). SHR1653 This task is definitely however very SHR1653 complex. First and foremost, the data currently available to perform the task are very scarce. Detailed information within the molecular relationships between an antibody and its cognate antigen target is currently available only from protein 3D constructions of antibodies co-crystallized with their target antigen, and currently the protein databank Mouse monoclonal to CD37.COPO reacts with CD37 (a.k.a. gp52-40 ), a 40-52 kDa molecule, which is strongly expressed on B cells from the pre-B cell sTage, but not on plasma cells. It is also present at low levels on some T cells, monocytes and granulocytes. CD37 is a stable marker for malignancies derived from mature B cells, such as B-CLL, HCL and all types of B-NHL. CD37 is involved in signal transduction (PDB) only consists of ~600 of such antibody-antigen (Ab-Ag) constructions. This scarceness makes it extremely complicated to learn rules of Ab-Ag relationships. Moreover, for these rules to be of practical use, we need the 3D constructions of the antibodies SHR1653 and antigens that we are investigating, that are in most cases not available. One can potentially predict such constructions with well-established methods (1114), but actually then it will be essential to address the effect that the accuracy of these models will have within the overall performance of any epitope prediction method. As a practical example, protein-protein molecular.