Drug side effects are becoming the leading factor of drug attrition in drug development and post-market drug withdrawal. Therefore, the prediction of potential drug side effects can help prevent the attrition of drug candidates and improve the efficiency of drugs. In this project, we aim to develop a novel method to predict potential side effects based on a heterogeneous network with integrating the molecular data and pharmacological data. We firstly propose a machine-learning-based approach for side effects related proteins prediction by integrating drug chemical structure and side effect similarity. Based on the integration of various of data, drugs can be described as a drug association network with each node denotes a drug and the edge can be described with their similarities, the side effects association network is constructed in the same manner. With two types of association networks, a heterogeneous network can be constructed by adding links representing drug-side effects associations across the two networks and deep learning method is applied to predict the new associations between drugs and side effects. Similarly, by investigating the associations between drug therapy and side effects, we also proposed a new random walk method to predict new indications for known drugs. The proposed method is expected to be useful in the drug development process.
药物副作用已成为药物研发损耗以及上市后召回的主要因素。因此提早预测药物副作用可以降低药物研发失败的风险,减少对人类健康的危害。近年来,大量多层次组学数据的不断涌现,为药物副作用的预测带来了新的机遇。本项目将利用这些多组学数据构建药物-副作用异构网络,并开发新的算法来预测药物未知的副作用。首先,基于药物的化学结构及其副作用的相关属性,开发新的机器学习的方法来预测药物副作用相关蛋白;其次,通过整合不同来源的数据构建药物关联网络、副作用关联网络和药物-副作用异构网络,并根据药物-副作用间的已知关联,利用深度学习的方法来预测药物潜在的副作用;最后,根据药物已有疗效与副作用间的关联规则,开发新的基于异构网络的随机游走算法来预测药物的新疗效。本项目的研究有助于理解药物副作用的发生机制,为新药研发提供便利。同时能够预测已有药物的未知疗效,为药物重定位提供新的思路。
近年来,大量多层次组学数据的不断涌现,为药物研发带来了新的机遇。本项目利用这些多组学数据构建药物相关异构网络,并开发新的算法来预测药物与疾病的关联关系。首先,基于药物的化学结构及其他相关属性,开发新的机器学习的方法来预测药物耐药机制相关蛋白;其次,通过整合不同来源的数据构建药物关联网络、疾病关联网络和药物-疾病异构网络;最后,根据药物已有疗效与副作用间及其疾病间的关联规则,开发新的基于异构网络的算法来预测药物的新疗效。本项目的研究有助于理解药物耐药及副作用的发生机制,为新药研发提供便利。同时能够预测已有药物的未知疗效,为药物重定位提供新的思路。
{{i.achievement_title}}
数据更新时间:2023-05-31
论大数据环境对情报学发展的影响
跨社交网络用户对齐技术综述
基于LASSO-SVMR模型城市生活需水量的预测
基于多模态信息特征融合的犯罪预测算法研究
面向云工作流安全的任务调度方法
“副作用-通路-候选药物”网络预测模型的建立及应用
基于代谢通路的药物毒副作用的预测研究
基于多尺度系统生物学的药物-副作用相互作用网络预测研究
基于异构网络的致病基因及其复合物预测方法研究