Onyekachi Nwogu, MS, 1. Division of Biomedical Informatics, Cincinnati Children's Hospital Medical Center; Cincinnati, Ohio, USA. 2. Department of Biomedical Informatics, University of Cincinnati, College of Medicine; Cincinnati, Ohio, USA. Kirandeep Gill, MS, 1. Division of Biomedical Informatics, Cincinnati Children's Hospital Medical Center; Cincinnati, Ohio, USA. 2. Department of Biomedical Informatics, University of Cincinnati, College of Medicine; Cincinnati, Ohio, USA. Carolina Moore, PhD, 1. Division of Biomedical Informatics, Cincinnati Children's Hospital Medical Center; Cincinnati, Ohio, USA. Krishna M. Roskin, PhD, 1. Division of Biomedical Informatics, Cincinnati Children's Hospital Medical Center; Cincinnati, Ohio, USA. 3. Division of Immunobiology, Cincinnati Children's Hospital Medical Center; Cincinnati, Ohio, USA. 4. Department of Pediatrics, University of Cincinnati, College of Medicine; Cincinnati, Ohio, USA.
Poster # 21
Introduction Convergent antibodies, antibodies elicited in multiple individuals in response to a common antigen, have been identified for dengue, HIV, and tetanus infection and influenza vaccination. Prior methods used to identify convergent antibodies use sequence similarity, however, antibodies with low sequence similarity can have similar structures and, consequently, similar function and affinity. We built a computational methodology that uses antibody structure information to identify convergent antibodies from adaptive immune receptor repertoire sequencing (AIRR-seq) data and used these to predict food allergen sensitization status. Our results show that using structure information greatly improves prediction accuracy. Methods The first and second complementary determining regions (CDR) were annotated by canonical class, a set of widely adopted canonical conformations. The third CDR was annotated by structural similarity to a database of solved antibody structures. Structurally convergent antibodies are defined as having matching CDR classification. The presence of structurally convergent antibodies, augmented with antibody isotype, were used for food allergen sensitization prediction using logistic regression. The allergic response depends on the crosslinking of IgE receptors on basophils and mast cells. We accounted for this immunobiology by considering combinations of IgE structural convergent antibodies in the model. Prediction accuracy was measured in the context of cross-validation. Results In this study, we used convergent antibodies to predict food allergen sensitization. We compared the use of sequence similarity with structural convergence. Models which use structural classification strongly outperform models using sequence similarity. Incorporating isotype into our structural model further improves performance. Our model, which utilizes combinations of IgE structures performs the best. Conclusion The use of antibody structural information greatly improves prediction accuracy in the context of food allergen sensitization. Incorporating knowledge of the immunobiology of food allergic disease into our methodology further improves our model performance.
Comentarios