AntiRef: reference clusters of human antibody sequences

Abstract

AbstractMotivationBiases in the human antibody repertoire result in publicly available antibody sequence datasets containing many duplicate or highly similar sequences. These redundant sequences are a barrier to rapid similarity searches and reduce the efficiency with which these datasets can be used to train statistical or machine learning models of human antibodies. Identity-based clustering provides a solution; however, the extremely large size of available antibody repertoire datasets makes such clustering operations computationally intensive and potentially out of reach for many scientists and researchers who would benefit from such data.ResultsAntiRef (Antibody Reference Clusters), which is modeled after UniRef, provides clustered datasets of filtered human antibody sequences. Due to the modular nature of recombined antibody genes, the clustering thresholds used by UniRef to cluster general protein sequences (100, 90, and 50 percent identity) are suboptimal for antibody clustering. Starting from a dataset of ∼451 million full-length, productive human antibody sequences from the Observed Antibody Space (OAS) repository, AntiRef provides antibody sequence datasets clustered at a range of identity thresholds better suited to antibody sequences. AntiRef90, which uses the least stringent clustering threshold, is roughly one-third the size of the input dataset and less than half the size of the non-redundant AntiRef100.AvailabilityAntiRef is freely available via Zenodo (zenodo.org/record/7474336) under the CC-BY 4.0 license, which matches the licensing of OAS datasets. All code used to download and filter sequences, generate AntiRef clusters, and create the figures used in this publication is freely available under the MIT license via GitHub (github.com/briney/antiref). We anticipate AntiRef updates will be released bi-annually, with the option for supplementary out-of-band updates when large or particularly interesting datasets are made available. The AntiRef versioning scheme (current version: v2022.12.14) refers to the date on which sequences were retrieved from OAS.Contactbriney@scripps.edu