RESUMO
Despite decades of antibody research, it remains challenging to predict the specificity of an antibody solely based on its sequence. Two major obstacles are the lack of appropriate models and the inaccessibility of datasets for model training. In this study, we curated >5,000 influenza hemagglutinin (HA) antibodies by mining research publications and patents, which revealed many distinct sequence features between antibodies to HA head and stem domains. We then leveraged this dataset to develop a lightweight memory B cell language model (mBLM) for sequence-based antibody specificity prediction. Model explainability analysis showed that mBLM could identify key sequence features of HA stem antibodies. Additionally, by applying mBLM to HA antibodies with unknown epitopes, we discovered and experimentally validated many HA stem antibodies. Overall, this study not only advances our molecular understanding of the antibody response to the influenza virus but also provides a valuable resource for applying deep learning to antibody research.
RESUMO
Antibody discovery is crucial for developing therapeutics and vaccines as well as understanding adaptive immunity. However, the lack of approaches to synthesize antibodies with defined sequences in a high-throughput manner represents a major bottleneck in antibody discovery. Here, we presented oPool+ display, which combines oligo pool synthesis and mRNA display to construct and characterize many natively paired antibodies in parallel. As a proof-of-concept, we applied oPool+ display to rapidly screen the binding activity of >300 natively paired influenza hemagglutinin (HA) antibodies against the conserved HA stem domain. Structural analysis of 16.ND.92, one of the identified HA stem antibodies, revealed a unique binding mode distinct from other known broadly neutralizing HA stem antibodies with convergent sequence features. Yet, despite such differences, 16.ND.92 remained broadly reactive and conferred in vivo protection. Overall, this study not only established an experimental platform that can be applied in both research and therapeutics to accelerate antibody discovery, but also provides molecular insights into antibody responses to the influenza HA stem, which is a major target for universal influenza vaccine development.
RESUMO
Despite decades of antibody research, it remains challenging to predict the specificity of an antibody solely based on its sequence. Two major obstacles are the lack of appropriate models and inaccessibility of datasets for model training. In this study, we curated a dataset of >5,000 influenza hemagglutinin (HA) antibodies by mining research publications and patents, which revealed many distinct sequence features between antibodies to HA head and stem domains. We then leveraged this dataset to develop a lightweight memory B cell language model (mBLM) for sequence-based antibody specificity prediction. Model explainability analysis showed that mBLM captured key sequence motifs of HA stem antibodies. Additionally, by applying mBLM to HA antibodies with unknown epitopes, we discovered and experimentally validated many HA stem antibodies. Overall, this study not only advances our molecular understanding of antibody response to influenza virus, but also provides an invaluable resource for applying deep learning to antibody research.
RESUMO
Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) infection has triggered the COVID-19 pandemic. Several factors induce hypoxia in COVID-19. Despite being hypoxic, some SARS-CoV-2-infected individuals do not experience any respiratory distress, a phenomenon termed 'silent (or happy) hypoxia'. Prolonged undetected hypoxia could be dangerous, sometimes leading to death. A few studies attempted to unravel what causes silent hypoxia, however, the exact mechanisms are still elusive. Here, we aim to understand how SARS-CoV-2 causes silent hypoxia.