Supervised fine-tuning of pre-trained antibody language models improves antigen specificity prediction

Antibodies play a crucial role in adaptive immune responses by determining B cell specificity to antigens and focusing immune function on target pathogens. Accurate prediction of antibody-antigen specificity directly from antibody sequencing data would be a great aid in understanding immune responses, guiding vaccine design, and developing antibody-based therapeutics. In this study, we present a method of supervised fine-tuning for antibody language models, which improves on previous results in binding specificity prediction to SARS-CoV-2 spike protein and influenza hemagglutinin. We perform supervised fine-tuning on four pre-trained antibody language models to predict specificity to these antigens and demonstrate that fine-tuned language model classifiers exhibit enhanced predictive accuracy compared to classifiers trained on pretrained model embeddings. The change of model attention activations after supervised fine-tuning suggested that this performance was driven by an increased model focus on the complementarity determining regions (CDRs). Application of the supervised fine-tuned models to BCR repertoire data demonstrated that these models could recognize the specific responses elicited by influenza and SARS-CoV-2 vaccination. Overall, our study highlights the benefits of supervised fine-tuning on pre-trained antibody language models as a mechanism to improve antigen specificity prediction. Author SummaryAntibodies are vigilant sentinels of our adaptive immune system that recognize and bind to targets on foreign pathogens, known as antigens. This interaction between antibody and antigen is highly specific, akin to a fitting lock and key mechanism, to ensure each antibody precisely targets its intended antigen. Recent advancements in language modeling have led to the development of antibody language model to decode specificity information in the sequences of antibodies. We introduce a method based on supervised fine-tuning, which enhances the accuracy of antibody language models in predicting antibody-antigen interactions. By training these models on large datasets of antibody sequences, we can better predict which antibodies will bind to important antigens such as those found on the surface of viruses like SARS-CoV-2 and influenza. Moreover, our study demonstrates the potential of the models to "read" B cell repertoire data and predict ongoing responses, offering new insights into how our bodies respond to vaccination. These findings have significant implications for vaccine design, as accurate prediction of antibody specificity can guide the development of more effective vaccines.