Comparison of different inputs.

Over the years, synergistic drug combinations therapies have attracted widespread attention due to its advantages of overcoming drug resistance, increasing treatment efficacy and decreasing toxicity. Compared to lengthy medical drugs experimental screening, mathematical models and algorithms show great potential in synergistic drug combinations prediction. In this paper, we introduce a novel mathematical algorithm, the Human Pathway Relationship Network Algorithm (HPRNA), which is designed to predict synergistic drug combinations for angina pectoris. We first reconstruct a novel angina pectoris drug dataset, which include drug name, drug metabolism, chemical formula, targets and pathways, then construct a comprehensive human pathway network based on the genetic similarity of the pathways which contain information about the targets. Finally, we introduce a novel indicator to calculate drug pair scores which measure the likelihood of forming synergistic drug combination. Experimental results on angina pectoris drug datasets convincingly demonstrate that the HPRNA makes efficient use of target and pathway information and is superior to previous algorithms.

多年来，协同药物联合疗法因可克服耐药性、提升治疗效果并降低毒性而受到广泛关注。相较于耗时冗长的药物实验筛选，数学模型与算法在协同药物组合预测领域展现出巨大应用潜力。本文提出一种全新的数学算法——人类通路关联网络算法（Human Pathway Relationship Network Algorithm，HPRNA），旨在预测针对心绞痛（angina pectoris）的协同药物组合。我们首先构建了一套全新的心绞痛药物数据集，涵盖药物名称、药物代谢特征、化学分子式、作用靶点及相关通路；随后基于包含靶点信息的通路的遗传相似性，搭建了一套完整的人类通路网络。最后，我们提出一种全新的指标用于计算药物对评分，以衡量其形成协同药物组合的可能性。在心绞痛药物数据集上开展的实验结果充分证明，HPRNA能够高效利用靶点与通路信息，且性能优于现有经典算法。