We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Quantitative Biology > Biomolecules

Title: Iterative Refinement Graph Neural Network for Antibody Sequence-Structure Co-design

Abstract: Antibodies are versatile proteins that bind to pathogens like viruses and stimulate the adaptive immune system. The specificity of antibody binding is determined by complementarity-determining regions (CDRs) at the tips of these Y-shaped proteins. In this paper, we propose a generative model to automatically design the CDRs of antibodies with enhanced binding specificity or neutralization capabilities. Previous generative approaches formulate protein design as a structure-conditioned sequence generation task, assuming the desired 3D structure is given a priori. In contrast, we propose to co-design the sequence and 3D structure of CDRs as graphs. Our model unravels a sequence autoregressively while iteratively refining its predicted global structure. The inferred structure in turn guides subsequent residue choices. For efficiency, we model the conditional dependence between residues inside and outside of a CDR in a coarse-grained manner. Our method achieves superior log-likelihood on the test set and outperforms previous baselines in designing antibodies capable of neutralizing the SARS-CoV-2 virus.
Comments: Accepted to ICLR 2022
Subjects: Biomolecules (q-bio.BM); Machine Learning (cs.LG)
Cite as: arXiv:2110.04624 [q-bio.BM]
  (or arXiv:2110.04624v3 [q-bio.BM] for this version)

Submission history

From: Wengong Jin [view email]
[v1] Sat, 9 Oct 2021 18:23:32 GMT (3954kb,D)
[v2] Fri, 15 Oct 2021 19:48:56 GMT (3954kb,D)
[v3] Thu, 27 Jan 2022 22:29:40 GMT (3956kb,D)

Link back to: arXiv, form interface, contact.