Glycoinformatics-assisted Protein and Antibody Epitopes Prediction Service
Unlock Precision in Vaccine and Therapeutic Design with CD BioGlyco's Advanced Epitope Prediction Services
A protein or antibody epitope is a specific region on the surface of an antigen that antibodies recognize and bind to. Epitopes can be linear, where all these amino acids belong to a sequential sequence, or conformational (discontinuous), i.e., involve larger segments of the protein molecule that are brought together as a result of the folding process in their 3D structure. This consists of computational processes that allow for the identification of these regions in a protein sequence, to be used during epitope prediction which assists vaccine design or diagnostic and therapeutic applications with antibodies.
Fig.1 Shotgun mutagenesis epitope mapping of antibodies against HER2 revealed a novel epitope. (Wikipedia)
CD BioGlyco's glycoinformatics-assisted protein and antibody epitopes prediction service uses computational methods to predict the epitopes of proteins and antibodies. Our service utilizes two main approaches: sequence-based and structure-based. Moreover, we also provide other
Glycoinformatics-assisted Analysis/Prediction Services for clients to address their research needs.
- Our sequence-based methods include linear epitope prediction and conformational epitope prediction. Based on amino acid composition, physicochemical properties, and sequence similarity linear epitope prediction is performed by Support Vector Machines (SVMs), Random Forests (RF) but also deep learning models. Conformational epitope predictions are trying to define a continuous patch that contains residues from the whole protein sequence but these patches also belong at once (locations) for intrinsic properties of proteins such as surface accessibility, solvent-excluded surfaces, and antigenic propensity and secondary structure combined in models like DiscoTope or PEPITO.
- The structure-based methods at CD BioGlyco include antibody-antigen docking and epitope prediction combined with docking, the former utilizes docking software such as HADDOCK and ZDOCK to predict antibody binding sites, and the latter combines sequence prediction results with docking results to improve accuracy.
- Our emerging methods such as AlphaFold2 extract epitope information by predicting the structure of antibody-antigen complexes. For example, PAbFold successfully predicted the epitope of mBG17 antibody binding to the SARS-CoV-2 nucleocapsid protein and validated the results by peptide competition ELISA experiments.

Publication
DOI: 10.1371/journal.pone.0217668
Technology: Deep learning models, AlphaFold2, Graph convolutional attention network (PECAN)
Journal: PloS one
Published: 2019
IF: 2.9
Results: This paper summarizes the progress made in antibody epitope prediction in recent years, and discusses the general features of antibody-antigen interactions and techniques for predicting epitopes. It is found that deep learning models are gradually surpassing traditional feature-based machine learning methods in epitope prediction. Nevertheless, sequence and structural features still provide important insights into the problem of antibody-antigen recognition. To improve prediction accuracy, this paper emphasizes the importance of combining sequence and structural information for epitope prediction. It introduces several emerging prediction methods, such as AlphaFold2 and PECAN.
Fig.2 K-TOPE determines epitopes by tiling proteins into k-mers. (Paull, et al., 2019)
Applications
- Our protein and antibody epitopes prediction service can be used to predict immunogenic regions to enhance the immune response.
- Our protein and antibody epitopes prediction service can be used to identify epitope regions on target antigens for the development of monoclonal antibodies.
- Our protein and antibody epitopes prediction service can be used in the development of diagnostic tools by identifying specific epitopes associated with diseases.
- Our protein and antibody epitopes prediction service can be used to understand autoimmune responses by mapping autoantibody epitopes.
Advantages
- We utilize advanced machine learning models, including deep learning, to achieve high prediction accuracy.
- We combine amino acid sequences and protein structural information for precise epitope mapping.
- We have high-throughput capabilities that allow for the screening of numerous antigens and antibodies simultaneously.
- We offer experimental validation services, including peptide competition ELISA, to confirm predicted epitopes.
Frequently Asked Questions (FAQs)
- What is glycoinformatics-assisted protein and antibody epitope prediction?
- Glycoinformatics-assisted epitope prediction involves using computational tools and models to identify specific regions on proteins (epitopes) that are recognized by antibodies. This service leverages glycoinformatics to integrate glycan-related data, enhancing the prediction accuracy for both linear and conformational epitopes.
- What makes CD BioGlyco's service stand out in epitope prediction?
- CD BioGlyco's service stands out due to its high accuracy, integration of sequence and structural data, rapid processing capabilities, and customization options. We employ state-of-the-art deep learning models and provide experimental validation support, ensuring reliable and robust predictions.
- Is the service customizable for specific research needs?
- Yes, CD BioGlyco provides tailored prediction services to meet specific research objectives. The models and methods can be adjusted based on the unique requirements of different projects, ensuring that researchers get the most relevant and accurate results.
At CD BioGlyco, our glycoinformatics-assisted protein and antibody epitopes prediction service significantly contributes to advancements in immunology, vaccine development, and therapeutic antibody research. Please feel free to contact us for more information about our
Structural and Functional Prediction Services!
References
- https://en.wikipedia.org/wiki/Epitope_mapping
- Zeng, X.; et al. Recent progress in antibody epitope prediction. Antibodies. 2023, 12(3): 52.
- DeRoo, J.; et al. PAbFold: linear antibody epitope prediction using AlphaFold2. bioRxiv. 2024.
- Paull, M.L.; et al. A general approach for predicting protein epitopes targeted by antibody repertoires using whole proteomes. PloS one. 2019, 14(9): e0217668.
For research use only. Not intended for any diagnostic use.
Related Services