AUTHOR=Miller Nathaniel L. , Clark Thomas , Raman Rahul , Sasisekharan Ram 

TITLE=Learned features of antibody-antigen binding affinity

JOURNAL=Frontiers in Molecular Biosciences

VOLUME=Volume 10 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/molecular-biosciences/articles/10.3389/fmolb.2023.1112738

DOI=10.3389/fmolb.2023.1112738

ISSN=2296-889X

ABSTRACT=Understanding the factors that govern the interactions between antibodies and their target antigen is of interest given that antibodies have proven to be hugely successful therapeutic options for a diverse range of diseases. Defining the predictors of binding affinity and specificity between antibody and antigen has been challenging owing to the huge diversity in the conformations of the complimentarity determining regions (CDRs) of antibodies and the mode of engagement between antibody and antigen. In this study using the affinity annotated three-dimensional structures of antibody-antigen complexes in SAbDAb database (https://opig.stats.ox.ac.uk/webapps/newsabdab/sabdab/), we extracted simple features and also applied additional feature-sets from previous learning models to classify antigen binding affinity of an antibody. We demonstrated that pairwise combinations of simple features involving CDR-L2 and total contacts between the antibody and antigen provide the best cross-validation AUC. The findings from this study set the stage for future studies aimed at optimization of antigen-binding of antibodies by improving features that show potential predictive value for binding affinity.