Agentic End-to-End De Novo Protein Design for Tailored Dynamics Using a Language Diffusion Model
Bo Ni, Markus J. Buehler
2025-02-17
Summary
This paper talks about VibeGen, a new AI system that can design proteins from scratch with specific movement patterns. It's like teaching a computer to be a molecular architect, creating tiny machines that move in exactly the ways scientists want.
What's the problem?
Proteins are important molecules that do many jobs in living things, and how they move is crucial to how they work. But it's really hard to design new proteins that move in specific ways because the relationship between a protein's building blocks (its sequence) and how it moves is very complicated.
What's the solution?
The researchers created VibeGen, an AI system that uses two main parts working together. One part designs protein sequences, and the other part checks if these designs will move the right way. VibeGen uses advanced AI techniques to create completely new proteins that aren't like anything found in nature, but still move exactly as intended. They tested their designs using detailed computer simulations to make sure the proteins really would behave as planned.
Why it matters?
This matters because it opens up new possibilities for creating custom proteins for all sorts of uses. Scientists could design better enzymes for making medicines or chemicals, create new materials with special properties, or make sensors that work in new ways. It's a big step towards being able to design molecular machines that can do almost anything we can imagine, which could lead to breakthroughs in medicine, materials science, and biotechnology.
Abstract
Proteins are dynamic molecular machines whose biological functions, spanning enzymatic catalysis, signal transduction, and structural adaptation, are intrinsically linked to their motions. Designing proteins with targeted dynamic properties, however, remains a challenge due to the complex, degenerate relationships between sequence, structure, and molecular motion. Here, we introduce VibeGen, a generative AI framework that enables end-to-end de novo protein design conditioned on normal mode vibrations. VibeGen employs an agentic dual-model architecture, comprising a protein designer that generates sequence candidates based on specified vibrational modes and a protein predictor that evaluates their dynamic accuracy. This approach synergizes diversity, accuracy, and novelty during the design process. Via full-atom molecular simulations as direct validation, we demonstrate that the designed proteins accurately reproduce the prescribed normal mode amplitudes across the backbone while adopting various stable, functionally relevant structures. Notably, generated sequences are de novo, exhibiting no significant similarity to natural proteins, thereby expanding the accessible protein space beyond evolutionary constraints. Our work integrates protein dynamics into generative protein design, and establishes a direct, bidirectional link between sequence and vibrational behavior, unlocking new pathways for engineering biomolecules with tailored dynamical and functional properties. This framework holds broad implications for the rational design of flexible enzymes, dynamic scaffolds, and biomaterials, paving the way toward dynamics-informed AI-driven protein engineering.