CholBindNet: Interpretable Neural Networks for Cholesterol Binding Site Prediction

Cholesterol is a key modulator of membrane protein structure and function, yet predicting cholesterol binding sites remains challenging due to its undrug-like physicochemical properties. Here, we curated more than 800 high-resolution transmembrane protein structures containing cholesterol and developed an interpretable, atom-based deep-learning framework, CholBindNet, comprising four model architectures: a 3D convolutional neural network, a graph neural network, a graph attention network, and a graph convolutional network. A Positive-Unlabeled (PU) training strategy was employed to address the scarcity of true negative samples resulting from the promiscuous nature of cholesterol binding. We show that CholBindNet substantially outperforms existing deep-learning models trained on general ligand-binding datasets. The performance and generalizability of the model were further demonstrated by rapidly assessing strong, median, and weak cholesterol-binding sites in the PIEZO2 ion channel in excellent agreement with computationally expensive all-atom molecular dynamics (MD) simulations. Additionally, strong model interpretability was achieved for CholBindNet through atom-level feature encoding, Grad-CAM visualization, and attention-based scoring analysis. Overall, CholBindNet provides an efficient and scalable approach for predicting cholesterol binding sites on membrane proteins, achieving performance comparable to MD simulations while offering mechanistic biophysical insights beyond amino-acid sequence. This work hence lays the foundation for future development of deep-learning models targeting membrane protein drug-binding sites and cholesterol-modulated therapeutics.