基于K-Means聚类的微生物群落结构研究

Study on Microbial Community Structure Based on K-Means Clustering

随着宏基因组学的不断发展,揭示了微生物菌群在研究中的重要作用。采用K-Means聚类算法对来源于北平顶猴阴道微生物群落OTUs数据集的27个样本进行研究,并与PCA主成分分析法进行对比。K-Means聚类将OTUs数据集分成4个Cluster,而PCA将OTUs数据集划分成5个Cluster。此外,结合样本的元数据-pH,发现样本间的pH值相似性更能与K-Means聚类的分类保持一致。相较于PCA主成分析方法,K-Means聚类能更精确地对OTUs数据集进行分类。

The development of macrogeome has shown that microbial flora plays an important role in the research and development of many aspects. A total of 27 samples from the OTUs data collection of the microbiological community of the North Mongolian monkey were studied by K Means clustering algorithm and compared with the PCA principal component analysis method. K-Means clustering divides the OTUs data set into four clusters. Interestingly, PCA divides the OTUs data set into five clusters. In addition, combining the sample metadata-pH, it is found that the pH similarity between the samples is more consistent with the classification of K-Means cluste ring. K-Means clustering classifies the OTUs data sets more accurately than the PCA principal analysis method.