Reproduce “InstructPro2: Natural Language and Biological Grounding Guided De Novo Protein Design” 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
Charles Novak, Zhenqiao Song, Chuan Li, Jianwen Xie, Lei Li
InstructPro2 generates protein sequences from natural-language instructions, with optional ligand or antigen context. It is pretrained on 205,727,986 annotation-grounded protein–description pairs and adapted using target encoders. On its pretraining benchmark, it reports global RMSD of 3.04 Å and functional-recovery F1 of 74.8, compared with Pinal’s 5.33 Å and 62.2; evaluations are computational, without wet-lab validation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.