Comparison of sequence- and structure-based antibody clustering approaches on simulated repertoire sequencing data

Understanding adaptive immune system responses is crucial for developing vaccines and therapies. High-throughput characterization of antibodies is now an integral part of immune profiling and drug discovery. One of the main challenges is to identify functionally related antibodies as these enable the formulation of more suitable antibodies for therapies, or aid in characterising immune responses in different individuals. By clustering similar antibody clones, functionally related antibodies can be identified.

Conventional antibody clustering approaches rely on sequence-based information, particularly CDRH3 sequence identity and V/J gene usage, to group antibodies. However, it is known that sequences of different clonal origin may lead to antibodies with similar binding properties, because they are similar in structure but not sequence.