R语言包：microeco:一个用于微生物群落生态学数据挖掘的R包：第五：trans_diff class

差异丰度检验是微生物群落数据分析的重要组成部分。它可以用来确定群落间差异的重要分类群。目前，trans_diff类有三种著名的方法来进行这种分析:metastat、LEfSe和random forest。这里我们介绍random forest 方法

> t1 <- trans_diff$new(dataset = dataset, 
+                      method = "rf", 
+                      group = "Group", 
+                      taxa_level = "Genus")
1243 input features ...
751 features are remained after removing unknown features ...
Start Kruskal-Wallis rank sum test for Group ...
432 taxa found significant ...
After P value adjustment, 397 taxa found significant ...
Taxa abundance table is stored in object$res_abund ...
rf analysis result is stored in object$res_diff ...
> g1 <- t1$plot_diff_bar(use_number = 1:20, 
+                        group_order = c("TW", "CW", "IW"))
> g1

#lefse方法

> t1 <- trans_diff$new(dataset = dataset, method = "lefse", group = "Group", alpha = 0.01, lefse_subgroup = NULL)

> t1$plot_diff_bar(use_number = 1:30,

width = 0.8,

group_order = c("CW", "IW", "TW")) +

ggsci::scale_color_npg() +

ggsci::scale_fill_npg()

> t1$res_diff[1:5, ]
                                                       Comparison                                                 Taxa Method Group      LDA      P.unadj        P.adj
k__Bacteria|p__Proteobacteria                        CW - IW - TW                        k__Bacteria|p__Proteobacteria  LEfSe    CW 4.845260 3.209570e-11 1.075904e-09
k__Bacteria|p__Acidobacteria                         CW - IW - TW                         k__Bacteria|p__Acidobacteria  LEfSe    IW 4.792228 5.749137e-12 2.955057e-10
k__Bacteria|p__Acidobacteria|c__Acidobacteria        CW - IW - TW        k__Bacteria|p__Acidobacteria|c__Acidobacteria  LEfSe    IW 4.791686 8.559155e-13 6.946430e-11
k__Bacteria|p__Bacteroidetes                         CW - IW - TW                         k__Bacteria|p__Bacteroidetes  LEfSe    TW 4.770984 1.190230e-09 1.529446e-08
k__Bacteria|p__Proteobacteria|c__Gammaproteobacteria CW - IW - TW k__Bacteria|p__Proteobacteria|c__Gammaproteobacteria  LEfSe    CW 4.624007 5.474697e-12 2.911029e-10
                                                     Significance
k__Bacteria|p__Proteobacteria                                 ***
k__Bacteria|p__Acidobacteria                                  ***
k__Bacteria|p__Acidobacteria|c__Acidobacteria                 ***
k__Bacteria|p__Bacteroidetes                                  ***
k__Bacteria|p__Proteobacteria|c__Gammaproteobacteria          ***

> t1$plot_diff_abund(use_number = 1:30)

然后，我们给出了分类树中差异特征的梯形图。这个数据集中的分类群太多了。

作为一个例子，我们只使用了树中最丰富的200个分类群和50个差异特征。

我们只在门级显示完整的分类标签，在其他级别使用字母来减少文本重叠。

#需要调用ggtree

library(ggtree)

t1$plot_diff_cladogram(use_taxa_num = 200,

use_feature_num = 50,

clade_label_level = 5,

group_order = c("CW", "IW", "TW"))