An ensemble 3D deep-learning model to predict protein metal-binding site

Predicting metal-binding sites in proteins is critical for understand-ing the protein's biological function. Here, we develop an ensemble deep convolutional neural network (CNN) method for predicting metal-binding sites based on their three-dimensional (3D) structure. We build multi-channel 3D voxels based on biophysical characteris-tics obtained from raw atom coordinates of each protein-binding pocket. Then, we use these 3D voxels as the input of an ensemble 3D CNN model. We train and evaluate the model using a curated da-taset of 3D protein structures. Our proposed model shows high per-formance in predicting metal-binding sites for Zn, Fe, Mg, Mn, Ca, and Na. Our approach offers a framework to use 3D spatial features to train 3D-CNN, which may be used to predict complicated metal -binding sites directly from their biophysical characteristics. The source code and webserver of the model are publicly available.

Automated summary

In this study, a neural network method based on three-dimensional structure was developed to predict metal-binding sites in proteins. By constructing multi-channel 3D voxels and training and evaluating the model, it showed high performance in predicting various metal-binding sites.