A comparison of antibody–antigen complex sequence‐to‐structure prediction methods and their systematic biases
Protein Science · 2024-09 · 33 (9)
DOI: 10.1002/pro.5127 ↗Kurz gefasst
Vergleichsstudie in der Zeitschrift Protein Science, online erschienen am 21.08.2024. Verfasst von Katherine Maia McCoy und 2 weiteren Autor:innen. Beteiligte Einrichtung: Molecular and Cell Biology Graduate Program, Dartmouth College, Hanover, New Hampshire, USA. Hauptthemen laut MEDLINE: Antigen-Antibody Complex. Die Arbeit wurde bisher 31-mal zitiert (Crossref) und nennt 50 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: NIH HHS, NIGMS NIH HHS, NCI NIH HHS.
Abstract (Originalsprache)
The ability to accurately predict antibody–antigen complex structures from their sequences could greatly advance our understanding of the immune system and would aid in the development of novel antibody therapeutics. There have been considerable recent advancements in predicting protein–protein interactions (PPIs) fueled by progress in machine learning (ML). To understand the current state of the field, we compare six representative methods for predicting antibody–antigen complexes from sequence, including two deep learning approaches trained to predict PPIs in general (AlphaFold‐Multimer and RoseTTAFold), two composite methods that initially predict antibody and antigen structures separately and dock them (using antibody‐mode ClusPro), local refinement in Rosetta (SnugDock) of globally docked poses from ClusPro, and a pipeline combining homology modeling with rigid‐body docking informed by ML‐based epitope and paratope prediction (AbAdapt). We find that AlphaFold‐Multimer outperformed other methods, although the absolute performance leaves considerable room for improvement. AlphaFold‐Multimer models of lower quality display significant structural biases at the level of tertiary motifs (TERMs) toward having fewer structural matches in non‐antibody‐containing structures from the Protein Data Bank (PDB). Specifically, better models exhibit more common PDB‐like TERMs at the antibody–antigen interface than worse ones. Importantly, the clear relationship between performance and the commonness of interfacial TERMs suggests that the scarcity of interfacial geometry data in the structural database may currently limit the application of ML to the prediction of antibody–antigen interactions.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Substanzen
Autor:innen und Institutionen
- Molecular and Cell Biology Graduate Program, Dartmouth College, Hanover, New Hampshire, USA.
- Thayer School of Engineering, Dartmouth College, Hanover, New Hampshire, USA.
- Department of Computer Science, Dartmouth College, Hanover, New Hampshire, USA.
ORCID-Profile: Katherine Maia McCoy, Gevorg Grigoryan
Förderung
- NIH HHS
- NIGMS NIH HHS
- NCI NIH HHS
- NIAID NIH HHS
- National Institutes of Health
- National Institute of Allergy and Infectious Diseases
Zitieren
APA
McCoy, K. M., Ackerman, M. E., & Grigoryan, G. (2024). A comparison of antibody–antigen complex sequence‐to‐structure prediction methods and their systematic biases. Protein Science, 33(9). https://doi.org/10.1002/pro.5127
Vancouver
McCoy KM, Ackerman ME, Grigoryan G. A comparison of antibody–antigen complex sequence‐to‐structure prediction methods and their systematic biases. Protein Science. 2024;33(9). doi:10.1002/pro.5127
BibTeX
@article{McCoy2024,
author = {McCoy, Katherine Maia and Ackerman, Margaret E and Grigoryan, Gevorg},
title = {A comparison of antibody–antigen complex sequence‐to‐structure prediction methods and their systematic biases},
journal = {Protein Science},
year = {2024},
volume = {33},
number = {9},
doi = {10.1002/pro.5127}
}
EUVIMED