Machine learning methods for deciphering the human immune repertoire language. Deciphering the immune information (immune status, antigen binding) encoded into antibody and T-cell repertoires is of paramount importance for the development of vaccines, diagnostics and therapeutics and requires machine learning approaches (artificial intelligence). We focus on developing deep learning approaches to learn how to read and write the immune repertoire language.
Single-cell immune repertoire and immuno-mass spectrometry methods. Our research focuses on developing experimental platforms that link droplet-based high-throughput single-cell sequencing with the large-scale mass-spectrometry analysis of serum antibody repertoires in order to resolve the diversity and specificity of the effector serum antibody repertoire at single antibody resolution.
Frank R, Widrich M, Akbar R, Klambauer G, Sandve GK, Robert PA, Greiff V(2026) Attribution assignment for deep-generative sequence models enables interpretability analysis using positive-only data PLoS Comput Biol, 22(9), e1014805 DOI 10.1371/journal.pcbi.1014805, PubMed 42804478
Zhong J, Cardente N, Sandve GK, Bashour H, Abbate MF, Greiff V(2026) PEPE: scalable extraction of multi-modal protein language model representations Bioinformatics, 42(6) DOI 10.1093/bioinformatics/btag375, PubMed 42286792
Stervbo U, Filippidis P, Breden F, Cowell LG, Davi F, Greiff V, Langerak AW, Luning Prak ET, Sharland AF, Miho E, Meysman P(2025) Challenges and future directions of AIRR-seq-based diagnostics Immunoinformatics (Amst), 19 DOI 10.1016/j.immuno.2025.100056, PubMed 42180390
