预测药物-靶点相互作用的异构网络嵌入模型研究

Research on heterogeneous network embedding model for predicting drug-target interactions

药物-靶点相互作用(DTIs)预测是药物发现的重要过程。随着计算技术的发展,基于生物数据的计算方法正高效率地加速这一过程。然而,这些方法大多忽略了靶标和药物的序列特征和异构性。本研究通过机器学习方法,提出多重网络嵌入框架(MLB-NEDTP)的预测模型,首先分析序列特征,然后将融合的特征嵌入多层异质信息网络中,以提高预测性能。多个数据集训练和验证结果表明,该模型与ATOMNET等最新模型相比具有明显优势。

Prediction of drug-target interactions(DTIs)is an important process in drug discovery.With the development of computing technology,biological data-based computing methods are accelerating this process efficiently.However,most of these methods ignore the sequence characteristics and heterogeneity of the target and drugs.In this study,a prediction model of machine learning based network embedding framework in multiplex networks(MLBNEDTP)which analyzes the sequence features and then embeds the merged features into a multi-layer heterogeneous information network to improve prediction performance.The model has been trained and validated by multiple data sets,whose results show that this model has obvious advantages over the latest models such as ATOMNET.