Reproduce “Variational Autoencoding Molecular Graphs with Denoising Diffusion Probabilistic Model” 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

Daiki Koge, Naoaki Ono, Shigehiko Kanaya

The authors combine a graph transformer, PIGVAE, and DDPM to represent molecules with probabilistic latent vectors linked by denoising diffusion. They pretrain on ZINC and fine-tune for property prediction using the initial latent vector. On held-out tests, the model has lower MSE than PIGVAE on all four datasets, including 0.2593 versus 0.3543 on CTSD; comparison does not isolate diffusion’s effect.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.