基于多组学数据融合的杜氏肌营养不良潜在药物预测方法研究

杜氏肌营养不良是一种X染色体隐性遗传疾病，全球平均每3500个新生男婴中就有一人罹患此病，而且目前医学界尚无有效疗法。基于多组学数据融合利用药物重定位策略进行杜氏肌营养不良潜在药物的预测，能够为杜氏肌营养不良的治疗提供有价值的指导和参考。因此，本项目拟从杜氏肌营养不良多组学数据出发，采用复杂网络模型进行数据融合，挖掘疾病高阶模式，设计预测杜氏肌营养不良潜在药物的高效算法。首先，基于杜氏肌营养不良多组学数据，构建多层网络模型，并研究网络的拓扑特性、鲁棒性、动态演化等。接着，基于图表示学习、张量理论、关联规则定义杜氏肌营养不良高阶模式、设计高效的多层网络高阶模式挖掘算法。最后，分析模式内部组织结构，探寻药物之间、疾病之间的复杂关联性、特异性，建立单个药物或者药物组合与杜氏肌营养不良之间的关系，发现治疗杜氏肌营养不良的潜在药物。

Duchenne muscular dystrophy is a recessive genetic disorder of the X chromosome, which affects one in every 3,500 newborn boys worldwide, and there is currently no effective treatment in the medical community. Based on multi-omics data fusion, using drug repositioning strategy to predict the potential drugs of Duchenne muscular dystrophy can provide valuable guidance and reference for the treatment of Duchenne muscular dystrophy. Therefore, this project intends to start from the multiomics data of Duchenne muscular dystrophy, use a complex network model for data fusion, mine higher-order patterns of disease, and design efficient algorithms for predicting Duchenne muscular dystrophy potential drugs. First, based on the multiomics data of Duchenne muscular dystrophy, a multilayer network model is constructed, and the topology characteristics, robustness, and dynamic evolution of the network are studied. Then, based on graph representation learning, tensor theory, and association rules, a high-order pattern of Duchenne muscular dystrophy is defined, and efficient multi-layer network high-order pattern mining algorithms are designed. Finally, we analyze the internal structure of the patterns, explore the complex correlations and specificities between drugs and diseases, and establish the relationship between a single drug or drug combination and diseases to discover potential drugs to treat Duchenne muscular dystrophy.