Reproduce “Atom-level Protein Representation Learning Improves Protein Structure Prediction” 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
Taewon Kim, Hyosoon Jang, Hyunjin Seo, Seonghwan Seo, Hyeongwoo Kim, Wonho Zhung, Mingyeong Shin, Wooyoun Kim and 1 more
TriProRep learns residue-level protein representations by combining amino-acid identity, backbone geometry, and local full-atom geometry, then training an encoder to recover these views after independent plausible corruptions. With the same 100M docking model, its 2.8B representation reaches DockQ 0.499 on RepSP homodimers, versus 0.448 for ESM3. This comparison does not isolate full-atom tokens as the cause.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.