Diffusion models for protein structure generation (and design)

Denoising Diffusion Probabilistic Models (DDPMs) are a class of generative models that can be used to create new data, such as new images or protein structures. This video presents image generation (like StableDiffusion or DALL-E) as a conceptual analogy to explore the protein diffusion models RF Diffusion, Chroma, and the pioneering Anand/Achim models. Noising and denoising, training for high-quality reconstruction are discussed.
Protein Modeling and Design with PyRosetta and Machine Learning
• Protein Modeling and D...
Video from the Rosetta Commons RaMP Bootcamp (July 2023)
Instructor: Deniz Akpinaroglu (UCSF)
Chroma paper: www.nature.com... paper: www.nature.com... model paper: arxiv.org/pdf/...
Credits:
Instructor: Deniz Akpinaroglu (UCSF)
RaMP Director and Rosetta Commons Director: Jeffrey Gray (JHU)
RaMP Program Administrator:Camille Mathis (JHU)
Rosetta Commons Instructional Designer: Ashley Vater (UC Davis)
Video Production: Elizabeth Bonilla (JHU)
Funding: Rosetta Commons, National Science Foundation, and Johns Hopkins University