A graph neural network model enables accurate prediction of Anaplastic lymphoma kinase inhibitors compared to other machine learning models

Abstract: Anaplastic lymphoma kinase (ALK), a tyrosine kinase receptor, is defined as an important target in the development of anticancer drugs for non-small cell lung cancer. In the present work, we have performed a graph neural network (GNN) model in comparison to three fingerprints machine learning (ML) models for rapid anticancer bioactivity prediction. ALK inhibitors with IC50 values were taken from the REAXYS database. After preprocessing, all these inhibitors resulted in a dataset of 1664 molecules. Then, GNN and ML models were built on a training set. The generalization of these models was assessed by internal and external validation procedures. The graph neural network model achieved promising results, with an average precision of 0.879±0.041 and an F1 score of 0.804±0.049 in cross-validation. In external validation, the model achieved an average precision of 0.938 and an F1 score of 0.863, outperforming other results of the ML models. We can conclude that the forecast model obtained by the graph neural network is suitable for the problem and can be employed to predict the biological activity of new ALK inhibitors.

Keywords: Anaplastic lymphoma kinase; Graph neural network; Deep learning model; Non-small cell lung cancer; Drug classification; Machine learning.

Link paper here

Recommended citation: T. -C. Trinh et al., “A Graph Neural Network Model Enables Accurate Prediction of Anaplastic Lymphoma Kinase Inhibitors Compared to Other Machine Learning Models,” 2023 15th International Conference on Knowledge and Systems Engineering (KSE), Hanoi, Vietnam, 2023, pp. 1-6, doi: 10.1109/KSE59128.2023.10299477.