采用偏最小二乘法的基因-药物共模块识别

Identify Gene-Drug Co-Modules by Partial Least Square Method

首先,将药物二维化学结构转化为数值序列,计算药物之间的皮尔逊相关系数,进而构建药物关联网络;然后,在带有基因网络约束的稀疏偏最小二乘算法的基础上,加入药物关联网络信息,提出伴有基因和药物关联网络正则约束的稀疏偏最小二乘(SGDPLS)算法;最后,将SGDPLS算法应用于基因-药物共模块识别.结果表明:药物关联网络信息的加入能够有效提高所识别的共模块中基因模块与药物模块的相关性,增加共模块的生物可解释性.

First,we transform the two-dimensional chemical structures of drugs into digital sequences,calculate the Pearson correlation coefficients between drugs,and then construct a drug association network.Next,we incorporate the information from drug association network into sparse partial least square algorithm with gene network,and presente the sparse partial least square algorithm with gene and drug association networks(SGDPLS)algorithm.Finally,we apply SGDPLS algorithm to identify gene-drug co-modules.The result shows that,the addition of drug association network can improve the correlations between the gene modules and drug modules identified from the common module,and enhance the interpretability of the modules.