S. Rosset

Accurate single domain scaffolding of three non-overlapping protein epitopes using deep learning

This paper designs small single domain proteins that present three distinct viral epitopes using deep learning, showing improved immunogenicity compared to single-epitope designs.

Dating ancient humans splits by estimating Poisson rates from mitochondrial DNA parity samples

The paper estimates divergence times between species using Poisson rate models on mitochondrial DNA parity data, showing improved accuracy over BEAST2 and suggesting later TMRCA for Neanderthals, Denisovans, and chimpanzees.

Rational design of chemically controlled antibodies and protein therapeutics

This study designs and tests small molecule-controlled switchable protein therapeutics by optimizing protein interactions and demonstrating in vitro and in vivo controllability with Venetoclax.

De novo design of site-specific protein interactions with learned surface fingerprints

This paper designs novel protein interactions using geometric deep learning to generate surface fingerprints, achieving nanomolar affinity for four targets through computational and experimental methods.

Assessing and enhancing foldability in designed proteins

This study develops methods to assess and improve foldability in designed proteins, significantly enhancing catalytic efficiency and stability.

A rational blueprint for the design of chemically-controlled protein switches

This paper presents a computational strategy to design chemically-controlled protein switches using drug-inhibited interactions, expanding chemical space for synthetic circuits in mammalian cells.

A bottom-up approach for the de novo design of functional proteins

This paper introduces a bottom-up method for designing functional proteins tailored to complex motifs, demonstrating their use as biosensors and signaling ligands.

De novo protein design enables precise induction of functional antibodies in vivo

This paper introduces a de novo protein design algorithm called TopoBuilder that generates immunogens inducing precise antibody responses against respiratory syncytial virus in mice and non-human primates.

ModelTeller: model selection for optimal phylogenetic reconstruction using machine learning

ModelTeller uses machine learning to select phylogenetic models for accurate branch-length estimation, outperforming traditional statistical methods on complex datasets.

Boosting subdominant neutralizing antibody responses with a computationally designed epitope-focused immunogen

This study demonstrates that a computationally designed epitope-focused immunogen enhances subdominant neutralizing antibody responses against RSV.