Metal3D: a general deep learning framework for accurate metal ion location prediction in proteins

• Published in: Nature communications Vol. 14; no. 1; p. 2713
• Main Authors: Dürr, Simon L., Levy, Andrea, Rothlisberger, Ursula
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 11.05.2023; Nature Publishing Group; Nature Portfolio
• Subjects: 631/114/1305; 631/114/469; 631/45/612/1141; 631/535/1267; Annotations; Artificial neural networks; Binding Sites; Deep Learning; Density; Design; Electronic structure; Extensibility; Heavy metals; Humanities and Social Sciences; Ions - chemistry; Machine learning; Metal ions; Metalloproteins - metabolism; Metals; Metals - chemistry; multidisciplinary; Neural networks; Predictions; Protein interaction; Proteins; Science; Science (multidisciplinary); Three dimensional models; Zinc; Zinc - metabolism
• DOI: 10.1038/s41467-023-37870-6
• ISSN: 2041-1723

Metal ions are essential cofactors for many proteins and play a crucial role in many applications such as enzyme design or design of protein-protein interactions because they are biologically abundant, tether to the protein using strong interactions, and have favorable catalytic properties. Computational design of metalloproteins is however hampered by the complex electronic structure of many biologically relevant metals such as zinc . In this work, we develop two tools - Metal3D (based on 3D convolutional neural networks) and Metal1D (solely based on geometric criteria) to improve the location prediction of zinc ions in protein structures. Comparison with other currently available tools shows that Metal3D is the most accurate zinc ion location predictor to date with predictions within 0.70 ± 0.64 Å of experimental locations. Metal3D outputs a confidence metric for each predicted site and works on proteins with few homologes in the protein data bank. Metal3D predicts a global zinc density that can be used for annotation of computationally predicted structures and a per residue zinc density that can be used in protein design workflows. Currently trained on zinc, the framework of Metal3D is readily extensible to other metals by modifying the training data. Zinc is an essential metal for many proteins. Here, the authors propose a model based on 3D convolutional networks to predict the location of zinc in experimental and computationally predicted structures within a framework readily extensible to other metals.