MOTIVATION: Allergenicity, like antigenicity and immunogenicity, is a property encoded linearly and non-linearly, and therefore the alignment-based approaches are not able to identify this property unambiguously. A novel alignment-free descriptor-based fingerprint approach is presented here and applied to identify allergens and non-allergens. The approach was implemented into a four step algorithm. Initially, the protein sequences are described by amino acid principal properties as hydrophobicity, size, relative abundance, helix and β-strand forming propensities. Then, the generated strings of different length are converted into vectors with equal length by auto- and cross-covariance (ACC). The vectors were transformed into binary fingerprints and compared in terms of Tanimoto coefficient. RESULTS: The approach was applied to a set of 2427 known allergens and 2427 non-allergens and identified correctly 88% of them with Matthews correlation coefficient of 0.759. The descriptor fingerprint approach presented here is universal. It could be applied for any classification problem in computational biology. The set of E-descriptors is able to capture the main structural and physicochemical properties of amino acids building the proteins. The ACC transformation overcomes the main problem in the alignment-based comparative studies arising from the different length of the aligned protein sequences. The conversion of protein ACC values into binary descriptor fingerprints allows similarity search and classification. AVAILABILITY AND IMPLEMENTATION: The algorithm described in the present study was implemented in a specially designed Web site, named AllergenFP (FP stands for FingerPrint). AllergenFP is written in Python, with GIU in HTML. It is freely accessible at http://ddg-pharmfac.net/Allergen FP. CONTACT: [email protected] or [email protected].

BACKGROUND: Allergy is a form of hypersensitivity to normally innocuous substances, such as dust, pollen, foods or drugs. Allergens are small antigens that commonly provoke an IgE antibody response. There are two types of bioinformatics-based allergen prediction. The first approach follows FAO/WHO Codex alimentarius guidelines and searches for sequence similarity. The second approach is based on identifying conserved allergenicity-related linear motifs. Both approaches assume that allergenicity is a linearly coded property. In the present study, we applied ACC pre-processing to sets of known allergens, developing alignment-independent models for allergen recognition based on the main chemical properties of amino acid sequences. RESULTS: A set of 684 food, 1,156 inhalant and 555 toxin allergens was collected from several databases. A set of non-allergens from the same species were selected to mirror the allergen set. The amino acids in the protein sequences were described by three z-descriptors (z1, z2 and z3) and by auto- and cross-covariance (ACC) transformation were converted into uniform vectors. Each protein was presented as a vector of 45 variables. Five machine learning methods for classification were applied in the study to derive models for allergen prediction. The methods were: discriminant analysis by partial least squares (DA-PLS), logistic regression (LR), decision tree (DT), naïve Bayes (NB) and k nearest neighbours (kNN). The best performing model was derived by kNN at k = 3. It was optimized, cross-validated and implemented in a server named AllerTOP, freely accessible at http://www.pharmfac.net/allertop. AllerTOP also predicts the most probable route of exposure. In comparison to other servers for allergen prediction, AllerTOP outperforms them with 94% sensitivity. CONCLUSIONS: AllerTOP is the first alignment-free server for in silico prediction of allergens based on the main physicochemical properties of proteins. Significantly, as well allergenicity AllerTOP is able to predict the route of allergen exposure: food, inhalant or toxin.

The exogenous proteins are processed by the host antigen-processing cells. Peptidic fragments of them are presented on the cell surface bound to the major hystocompatibility complex (MHC) molecules class II and recognized by the CD4+ T lymphocytes. The MHC binding is considered as the crucial prerequisite for T-cell recognition. Only peptides able to form stable complexes with the MHC proteins are recognized by the T-cells. These peptides are known as T-cell epitopes. All T-cell epitopes are MHC binders, but not all MHC binders are T-cell epitopes. The T-cell epitope prediction is one of the main priorities of immunoinformatics. In the present study, three chemometric techniques are combined to derive a model for in silico prediction of peptide binding to the human MHC class II protein HLA-DP1. The structures of a set of known peptide binders are described by amino acid z-descriptors. Data are processed by an iterative self-consisted algorithm using the method of partial least squares, and a quantitative matrix (QM) for peptide binding prediction to HLA-DP1 is derived. The QM is validated by two sets of proteins and showed an average accuracy of 86 percent.

MOTIVATION: T-cell epitope identification is a critical immunoinformatic problem within vaccine design. To be an epitope, a peptide must bind an MHC protein. RESULTS: Here, we present EpiTOP, the first server predicting MHC class II binding based on proteochemometrics, a QSAR approach for ligands binding to several related proteins. EpiTOP uses a quantitative matrix to predict binding to 12 HLA-DRB1 alleles. It identifies 89% of known epitopes within the top 20% of predicted binders, reducing laboratory labour, materials and time by 80%. EpiTOP is easy to use, gives comprehensive quantitative predictions and will be expanded and updated with new quantitative matrices over time. AVAILABILITY: EpiTOP is freely accessible at http://www.pharmfac.net/EpiTOP.