Torben Schiffner, Ph.D.

A central focus of the Schiffner Lab is the development of computational tools that accelerate and improve the design of vaccine immunogens. While modern deep-learning methods have transformed protein design, no current algorithm can reliably predict protein–protein interaction affinities — a key requirement for designing immunogens that selectively engage rare antibody precursors. The lab is generating high-throughput affinity datasets using yeast-display technologies and using these data to train new neural networks, including message-passing neural networks (MPNNs) and fine-tuned protein language models, that combine sequence and structural information. By integrating these emerging methods with established physics-based design platforms such as Rosetta, the lab aims to substantially increase the success rate of computational immunogen design.

The Schiffner Lab continues a long-standing program in HIV-1 vaccine development built around germline targeting, the strategy of designing immunogens that selectively activate the unmutated B-cell precursors of broadly neutralizing antibodies. Schiffner led the development of 10E8-GT12 24mer, a germline-targeting epitope-scaffold immunogen that activates precursors of the HIV-1 bnAb 10E8 in preclinical models and engages corresponding precursors in human blood (Schiffner et al., Nat Immunol, 2024; Ray et al., Nat Immunol, 2024). Building on these advances, the lab is developing additional epitope scaffolds and heterologous boosting immunogens needed to shepherd antibody affinity maturation toward broad neutralization. Recent advances in the lab’s computational pipeline have achieved >50% success rates in de novo epitope-scaffold design, enabling the rapid generation of multiple diverse immunogens for the same antibody lineage.

As part of the ARPA-H–funded “Protect against Emergent Alphaviruses through Computation” (PEAC) consortium, a multi-institutional collaboration combining 14 principal investigators from seven institutions, the Schiffner Lab strives to design AI/ML-based vaccines that broadly protect against diverse alphaviruses. Unlike HIV-1 bnAbs, which are highly mutated, broadly neutralizing antibodies against alphaviruses carry relatively few somatic mutations, opening the possibility of priming and boosting bnAbs with a single vaccine formulation. The lab is developing combined prime-boost cocktails and heterologous prime-boost nanoparticle platforms tailored to lightly mutated bnAb lineages.

Through an NIH U01 award, the Schiffner Lab is applying its AI-enabled immunogen design pipeline to influenza A virus, with the goal of developing immunogens that engage broadly neutralizing antibody precursors against conserved epitopes on hemagglutinin. The lab is also extending its work on pre-fusion stabilization, glycan masking, and germline targeting to additional pathogens including hepatitis C virus and emerging coronaviruses.