作者:贾继华，许耀奎，王明辉

Authors:JIA Jihua,XU Yaokui,WANG Minghui
摘要:近年来单细胞RNA 测序(scRNA-seq)技术的快速发展使得在单个细胞水平上研究组织器官的异质性成为可能 。针对单细胞RNA 测序数据中准确鉴定细胞类型问题,提出一种新的基于双自编码结合变分贝叶斯高斯混合模型的聚类方法,称之为 sc-VBDAE。首先通过对抗自编码网络的编码和解码过程重构数据,然后使用经典 自编码对数据进行降维,获得低维且有效的数据 。最后使用变分贝叶斯高斯混合模型对细胞进行聚类 ,并可视化聚类结果 。在 10 个 scRNA-seq 数据上的实验结果表明,该方法在 6 个数据集上 ARI 指标均优于其它方法 ,在数据 集 Biase 和 Klein 上 ARI 指标值达到 0. 90 及以上。
Abstract:In recent years, the rapid development of single-cell RNA sequencing(scRNA-seq) technology has made it possible to research the heterogeneity of tissues and organs at the single-cell level. To accurately identify cell types in scRNA-seq data, based on dual autoencoder combined with variational Bayesian Gaussian mixture mode, a new clustering method, sc-VBDAE, is proposed. First, through the encoding and decoding process in adversarial autoencoder network, the scRNA-seq data is reconstructed. Then, the autoencoder network is used to reduce the dimensionality of the data, so as to obtain low-dimensional and effective scRNA-seq data. Finally, the variational Bayesian Gaussian mixture model is used to cluster the cells and visualize the clustering results. The experimental results on ten scRNA-seq datasets show that the ARI index of the proposed method is superior to other methods on six datasets, and the ARI index value on Biase and Klein datasets reaches 0. 90 or above.

PDF全文下载地址:

可免费Download/下载PDF全文