New AI model reconstructs molecules from coarse-grained representations

Researchers have developed a new method for compositional backmapping, which reconstructs molecules from coarse-grained (CG) representations. This technique uses a discrete denoising diffusion model called Juniper, conditioned on the octanol-water partition free energy ($\Delta G_{\mathrm{W} \mapsto \mathrm{O}}$). The model, trained on molecules mapped to one or two beads, generates valid and unique molecules with distributions that closely track the target free energy. This advancement allows CG screening results to be translated into candidate molecules for further study or synthesis. AI

IMPACT Enables the translation of coarse-grained simulation results into specific molecular candidates for synthesis or atomistic study.

RANK_REASON The cluster contains an arXiv preprint detailing a new AI model for a scientific research problem. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• $\Delta G_{\mathrm{W} \mapsto \mathrm{O}}$
• Hugging Face
• Luis Itza Vazquez-Salazar
• Martini

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →