Researchers have developed a new computational framework, Gen-COMPAS, to simulate complex molecular transitions like protein folding and membrane transport. The method addresses limitations of existing techniques, which are often computationally expensive and prone to bias.
Conventional molecular dynamics simulations struggle with the rarity of these transitions. Enhanced-sampling strategies, while attempting to overcome this, require significant computational resources and rely on potentially arbitrary parameters. Gen-COMPAS, introduced by Tangcyu and colleagues, reconstructs transition pathways without predefined collective variables, offering a more efficient approach.
The framework combines a denoising diffusion probabilistic model, generating plausible intermediate structures, with committor-based filtering to pinpoint transition states. Short, unbiased simulations from these intermediates achieve sampling scales that would require far greater computational effort with traditional methods. The researchers demonstrated Gen-COMPAS’s effectiveness on systems ranging from a miniprotein to a pentameric, ligand-gated ion channel.
According to the study, Gen-COMPAS successfully recovers committors, transition states, and free-energy landscapes using only known end-point structures, eliminating the need for prior mechanistic knowledge or predefined reaction coordinates. The source data and code for Gen-COMPAS are available on GitHub (https://github.com/Tangcyu/Gen-COMPAS).
閱讀原始報道
💬 评论
📜 评论政策