Reproduce “Designing Cyclic Peptides via Harmonic SDE with Atom-Bond Modeling” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xiangxin Zhou, Mingyu Li, Yi Xiao, Jiahan Li, Dongyu Xue, Zaixiang Zheng, Jianzhu Ma, Quanquan Gu
Researchers developed CPSDE, a generative framework that directly designs diverse cyclic peptides guided by a protein target by explicitly modeling all atoms and bonds. The framework consistently generates cyclic peptides with superior stability, affinity, and structural diversity compared to linear peptide design methods, as demonstrated by quantitative metrics and molecular dynamics simulations.
Key insights
- Explicit all-atom and bond modeling in ATOMSDE enables generalization across small molecules, linear, and diverse cyclic peptides, effectively addressing data scarcity.
- The harmonic Stochastic Differential Equation (SDE) implicitly incorporates bond constraints from the outset, providing a structured prior that guides the generative process for valid molecular structures.
- A routed sampling algorithm iteratively updates both sequence and structure through alternating ATOMSDE (structure denoising) and RESROUTER (sequence prediction) steps, allowing for dynamic chemical graph updates critical for complex cyclic peptide formation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.