Researchers have developed Harmony, a new framework for protein-ligand flexible docking that explicitly accounts for the periodic geometry of angular variables. Unlike previous diffusion-based models that use generic Euclidean heads, Harmony parameterizes torsional score fields using learned harmonic potentials on the circle. This approach incorporates a frequency-aware inductive bias for rotameric motion, leading to improved ligand pose accuracy and pocket reconstruction on the PDBBind benchmark. Harmony also enhances the physical validity of generated complexes on the PoseBusters dataset and has been demonstrated on specific binding sites like EBNA1 and KRAS G12D. AI
IMPACT This research could lead to more accurate and physically valid predictions in drug discovery and molecular biology simulations.
RANK_REASON The cluster contains a research paper detailing a new computational method for molecular docking. [lever_c_demoted from research: ic=1 ai=1.0]
- arXiv
- DNA-binding domain
- Harmony
- KRAS G12D
- PDBbind database
- PoseBusters: AI-based docking methods fail to generate physically valid poses or generalise to novel sequences
- protein family
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
