差异分析

#
# ### 这个函数输入数据框格式
# #这里备注所需的数据格式
# #前量列从第一列开始是ID，第二列是分组信息，剩下的列均为数据列
#
#
# single_diff_and_plot (data_wt,plotname = "南京农业大学_",plot = "bar",method_Mc ="LSD")
# single_diff_and_plot (data_wt,plotname = "南京农业大学_",plot = "bar",method_Mc ="scheffe")
# single_diff_and_plot (data_wt,plotname = "资源与环境科学学院_",plot = "box",method_Mc ="LSD" )
# single_diff_and_plot (data_wt,plotname = "资源与环境科学学院_",plot = "box",method_Mc = "scheffe" )
# method_Mc == "Duncan"
# single_diff_and_plot (data_wt,plotname = "资源与环境科学学院_",plot = "box",method_Mc = "Duncan" )


#method_cv: 方差齐心方法选择,有两种方法：method_cv == "leveneTest",method_cv == "bartlett.test"
#method_Mc:多重比较方法选择：默认Tukey。可选：LSD，
# i = 3

##更新

#我将方差分析的五种多重比较方法添加到函数中，共判断选用
#非参数检验两两检验，本来柱状图无法转化成字母标记，我将其转化为字母标记并使用。

#
# single_diff_and_plot (data_wt,plotname = "资源与环境科学学院_",plot = "box",method_Mc = "Duncan",sig_show = "abc" )
#
# single_diff_and_plot (data_wt,plotname = "资源与环境科学学院_",plot = "box",method_Mc = "Duncan",sig_show = "line" )
# single_diff_and_plot (data_wt,plotname = "资源与环境科学学院_",plot = "box",method_Mc = "Duncan",sig_show = "abc" )
#
#
#
#
# single_diff_and_plot (data_wt,plotname = "资源与环境科学学院_",plot = "bar",method_Mc = "Duncan",sig_show = "line")
# library(easyMicrobiome)
#
# data("data_wt")
# single_diff_and_plot (data_wt,plotname = "资源与环境科学学院_",plot = "bar",method_Mc = "Duncan",sig_show = "abc")
# #
# head(data_wt)
#
#
# sig_show = "line"
# sig_show = "abc"
# plotname = "wentao_"
# plot = "bar"
# method_cv = "bartlett.test"
# method_Mc = "Tukey"
#
# i = 1


# ## 贪婪模式：
# Greedy_model = "aov"
#
# Greedy_model = "non-parametric"

library(tidyverse)
library(agricolae)
# library(car)
library(reshape2)
library("ggplot2")
library("ggpubr")
library("ggsignif")
#
single_diff_and_plot <- function(data_wt = data_wt,plotname = "wentao_",plot = "bar",
                                 method_cv = "bartlett.test",method_Mc = "Tukey",sig_show = "abc")  {

Mytheme <- theme_bw()+

  # scale_fill_manual(values = mi, guide = guide_legend(title = NULL))+
  theme(

    panel.grid.major=element_blank(),
    panel.grid.minor=element_blank(),

    plot.title = element_text(vjust = -8.5,hjust = 0.1),
    axis.title.y =element_text(size = 20,face = "bold",colour = "black"),
    axis.title.x =element_text(size = 24,face = "bold",colour = "black"),
    axis.text = element_text(size = 20,face = "bold"),
    axis.text.x = element_text(colour = "black",size = 14),
    axis.text.y = element_text(colour = "black",size = 14),
    legend.text = element_text(size = 15,face = "bold"),
    legend.position = "none"#是否删除图例

  )
  # print(i)

  for (i in 3:ncol(data_wt)) {
    #构造待分析的子数据框
    ss <- data_wt[i]
    colnames(ss) <- c("count")
    ss$group = data_wt$group

    #正态检验
    xx <-  shapiro.test(as.vector(as.matrix(data_wt[i])))
    (p1 <- xx[[2]])#得要p值
    p1 <- round(p1,3)#保留三为小数

    #方差齐性检验
    if (method_cv == "leveneTest" ) {
      library("car")
      xc <- leveneTest(count~group,data=ss)
      (p2 <- xc[[3]][1])

    }


    if (method_cv == "bartlett.test" ) {

      xc <- bartlett.test(count~group,data=ss)
      (p2 <- xc[[3]])
      p2 <- round(p2,3)


    }

    if ( plot == "bar") {
      name_i = colnames(data_wt[i])
      #求取均值和方差
      wen1 = as.data.frame(tapply(as.vector(as.matrix(data_wt[i])),data_wt$group,mean,na.rm=TRUE))
      wen2 = as.data.frame(tapply(as.vector(as.matrix(data_wt[i])),data_wt$group,sd,na.rm=TRUE))
      went = cbind(wen1,wen2)

      colnames(went) = c("mean" ,"SD")
      went
      #p1 >=.05& p2 >=.05:数据符合正态分布，方差齐心
      if (p1 >=.05& p2 >=.05) {
        #进行方差检验 下面wtx3为提取的p值
        model<-aov(count ~ group, data= ss)#方差分析
        wtx1 = summary(model)
        wtx2 = wtx1[[1]]
        wtx3 = wtx2[5]#

        #wtx3$`Pr(>F)`[1]< 0.05：当P值小于0.05时，进行多重比较
        if ( wtx3$`Pr(>F)`[1]< 0.05) {

          if (method_Mc == "Tukey") {
            library(multcomp)

            litter.mc <- glht(model, linfct = mcp(group = 'Tukey'))
            # summary(litter.mc)
            insx = cld(litter.mc)
            aa <- insx$mcletters$monospacedLetters
            aa = as.data.frame(aa)
            colnames(aa) = c("groups")
            head(aa)
            aa$group = row.names(aa)

          }


          if (method_Mc == "LSD") {
            out <- LSD.test(model,"group", p.adj="none")#进行多重比较，不矫正P值
            aa = out$group#结果显示：标记字母法
            aa$group = row.names(aa)
            aa
          }
          #【SNK法】（Student-Newman-Keuls）程序运行结果与LSD.test类似。
          if (method_Mc == "SNK") {
            library("agricolae")
            out <- SNK.test(model,"group")
            aa = out$groups# 查看每个组的label

            aa$group = row.names(aa)
            stat = aa
            aa
          }

          #【Duncan法】(新复极差法)（SSR）
          if (method_Mc == "Duncan") {
            # library("agricolae")
            out <-duncan.test (model,"group")
            aa = out$groups# 查看每个组的label

            aa$group = row.names(aa)
            stat = aa
            aa
          }

          #Scheffe检验特点：各组样本数相等或不等均可以，但是以各组样本数不相等使用较多；
          if (method_Mc == "scheffe") {
            # library("agricolae")
            out <-scheffe.test (model,"group")
            aa = out$groups# 查看每个组的label

            aa$group = row.names(aa)
            stat = aa
            aa
          }


          wentao = merge(aa,went, by="row.names",all=F)
          wentao
          FileName <- paste(plotname ,name_i,method_Mc,"_aov_bar", ".csv", sep = "_")
          write.csv(wentao,FileName,quote = F)
          library(tidyverse)
          # colnames(wentao) = c(colnames(wentao[1:4]),"mean" ,"SD")
          #使用的tidyverse函数，对数据框添加两列，目的为柱状图添加bar
          aa = mutate(wentao, ymin = mean - SD, ymax =  mean + SD)
          a = max(aa$mean)*1.5##用于设置y轴最大值

          ### 出图柱状图
          p = ggplot(aa , aes(x = group, y = mean,colour= group)) +
            geom_bar(aes(colour= group,fill = group),stat = "identity", width = 0.4,position = "dodge") +

            geom_errorbar(aes(ymin=ymin,
                              ymax=ymax),
                          colour="black",width=0.1,size = 1)+
            # geom_hline(aes(yintercept=mean(as.vector(as.matrix(data_wt[i])))), colour="black", linetype=2) +
            # geom_vline(aes(xintercept=0), colour="black", linetype="dashed") +
            scale_y_continuous(expand = c(0,0),limits = c(0,a))+
            labs(x=paste(name_i,"of all group", sep = "_"),
                y="group",
                    title = paste("Normality test",p1,"Homogeneity of variance",p2,sep = ":"))
          p
          head(aa)
          if (sig_show == "line") {
            zuhe = combn(aa$group,2)
            xxxx <- tapply(zuhe,rep(1:ncol(zuhe),each=nrow(zuhe)),function(i)i)
            xxxx
            sig_lis = rep("a",dim(zuhe)[2])
            for (i in 1:dim(zuhe)[2]) {
              library(tidyverse)
              library("ggsignif")

              if (filter(aa, group == xxxx[[i]][1])$groups == filter(aa, group == xxxx[[i]][2])$groups) {
                sig_lis[i] = "no_sig"
              }

              if (filter(aa, group == xxxx[[i]][1])$groups != filter(aa, group == xxxx[[i]][2])$groups) {
                sig_lis[i] = "*"
              }

            }

           p = p +
              geom_signif(comparisons = xxxx, annotations=sig_lis,
                          y_position = (seq(from=1, to=max(aa$mean)/4,length.out=dim(zuhe)[2]) + max(aa$mean)), tip_length = rep(0.03,dim(zuhe)[2]),color = "black")
            p
          }




          if (sig_show == "abc") {


            p = p + geom_text(aes(label = groups,y=ymax, x = group,vjust = -0.3,size = 6))
            p
          }
          #as.vector(as.matrix(data_wt[i]))为进行差异分析的一组数据
          p=p+Mytheme
          p

          if (length(unique(data_wt$group))>3){	p=p+theme(axis.text.x=element_text(angle=45,vjust=1, hjust=1))}
          p
          FileName <- paste(plotname ,name_i,method_Mc,"_aov_bar", ".pdf", sep = "_")
          ggsave(FileName, p, width = 8, height = 8)

          #wtx3$`Pr(>F)`[1]>= 0.05当p值大于0.05那么就不能做多重比较了
        }else if ( wtx3$`Pr(>F)`[1]>= 0.05)  {

          colnames(went) = c("mean" ,"SD")
          aa = mutate(went, ymin = mean - SD, ymax =  mean + SD)
          aa$group = levels(ss$group)
          aa
          a = max(aa$mean)*2
          FileName <- paste(plotname ,name_i,method_Mc,"_aov_bar", ".csv", sep = "_")
          write.csv(aa,FileName,quote = F)
          p = ggplot(aa , aes(x = group, y = mean,colour= group)) +
            geom_bar(aes(colour= group,fill = group),stat = "identity", width = 0.4,position = "dodge") +
            # geom_text(aes(label = groups,y=ymax, x = group,vjust = -0.3,size = 6))+
            geom_errorbar(aes(ymin=ymin,
                              ymax=ymax),
                          colour="black",width=0.1,size = 1)+
            scale_y_continuous(expand = c(0,0),limits = c(0,a))+
            labs(x=paste(name_i,"of all group", sep = "_"),
                 y="group",
                 title = paste("Normality test",p1,"Homogeneity of variance",p2,"aov",round(wtx3$`Pr(>F)`[1],3),sep = ":"))
          p
          p=p+Mytheme
          p
          if (length(unique(data_wt$group))>3){	p=p+theme(axis.text.x=element_text(angle=45,vjust=1, hjust=1))}
          FileName <- paste(plotname ,name_i,"aov_nosig"," bar", ".pdf", sep = "_")
          ggsave(FileName, p, width = 8, height = 8)
        }


        #   p1 <.05| p2 <.05：不符合正态检验或者方差不齐性
      }else if (p1 <.05| p2 <.05 ){
        library(ggpubr)
        library("multcompView")
        i
        krusk=compare_means(count ~ group, data=ss, method = "kruskal.test")
        krusk
        sumkrusk=as.data.frame(krusk)
        sumkrusk
        #多组比较小于0.05，表明多组之间具有差异，可以进行两两非参数检验，并标记字母
        #但是这里没有做，个人认为挺难的，还
        if ( sumkrusk[3]< 0.05) {

          krusk=compare_means(count ~ group, data=ss, method = "wilcox.test")

          xx=as.data.frame(krusk)
          xx$group1
          wilcox_levels = paste(xx$group1,xx$group2,sep = "-")
          wilcox_levels = xx$p


          names(wilcox_levels) =  paste(xx$group1,xx$group2,sep = "-")

          wilcox.labels <- data.frame(multcompLetters(wilcox_levels, threshold = 0.05)['Letters'])
          colnames(wilcox.labels) = "groups"
          aa = wilcox.labels
          wentao = merge(aa,went, by="row.names",all=F)
          wentao

          # colnames(went) = c("mean" ,"SD")
          aa = mutate(wentao, ymin = mean - SD, ymax =  mean + SD)
          aa$group = levels(ss$group)
          aa
          FileName <- paste(plotname ,name_i,"_wilcox.test_YES_bar", ".csv", sep = "_")
          write.csv(aa,FileName,quote = F)
          a = max(aa$mean)*1.5

          ### 出图柱状图
          p = ggplot(aa , aes(x = group, y = mean,colour= group)) +
            geom_bar(aes(colour= group,fill = group),stat = "identity", width = 0.4,position = "dodge") +

            geom_errorbar(aes(ymin=ymin,
                              ymax=ymax),
                          colour="black",width=0.1,size = 1)+
            # geom_hline(aes(yintercept=mean(as.vector(as.matrix(data_wt[i])))), colour="black", linetype=2) +
            # geom_vline(aes(xintercept=0), colour="black", linetype="dashed") +
            scale_y_continuous(expand = c(0,0),limits = c(0,a))+
            labs(x=paste(name_i,"of all group", sep = "_"),
                 y="group",
                 title = paste("Normality test",p1,"Homogeneity of variance",p2,sep = ":"))
          p
          head(aa)
          if (sig_show == "abc") {
            p = p +
              geom_text(aes(label = groups,y=ymax, x = group,vjust = -0.3,size = 6))
            p
          }

          if (sig_show == "line") {
            zuhe = combn(aa$group,2)
            xxxx <- tapply(zuhe,rep(1:ncol(zuhe),each=nrow(zuhe)),function(i)i)
            xxxx
            sig_lis = rep("a",dim(zuhe)[2])
            for (i in 1:dim(zuhe)[2]) {
              library(tidyverse)

              if (filter(aa, group == xxxx[[i]][1])$groups == filter(aa, group == xxxx[[i]][2])$groups) {
                sig_lis[i] = "no_sig"
              }

              if (filter(aa, group == xxxx[[i]][1])$groups != filter(aa, group == xxxx[[i]][2])$groups) {
                sig_lis[i] = "*"
              }

            }

            p = p +
              geom_signif(comparisons = xxxx, annotations=sig_lis,
                          y_position = (seq(from=1, to=max(aa$mean)/4,length.out=dim(zuhe)[2]) + max(aa$mean)), tip_length = rep(0.03,dim(zuhe)[2]),color = "black")
            p
          }



          #as.vector(as.matrix(data_wt[i]))为进行差异分析的一组数据
          p=p+Mytheme
          p
          if (length(unique(data_wt$group))>3){	p=p+theme(axis.text.x=element_text(angle=45,vjust=1, hjust=1))}
          FileName <- paste(plotname ,name_i,"_kruskal.test_YES_bar", ".pdf", sep = "_")
          ggsave(FileName, p, width = 8, height = 8)
          # 如果多组比较大于0.05表明多组之间不具有差异，那么直接出图
        }else if ( sumkrusk[3] >= 0.05)  {
          colnames(went) = c("mean" ,"SD")
          aa = mutate(went, ymin = mean - SD, ymax =  mean + SD)
          aa$group = levels(ss$group)
          aa
          a = max(aa$mean)*1.5
          FileName <- paste(plotname ,name_i,"_kruskal.test_nosig_bar", ".csv", sep = "_")
          write.csv(aa,FileName,quote = F)
          p = ggplot(aa , aes(x = group, y = mean,colour= group)) +
            geom_bar(aes(colour= group,fill = group),stat = "identity", width = 0.4,position = "dodge") +
            geom_errorbar(aes(ymin=ymin,
                              ymax=ymax),
                          colour="black",width=0.1,size = 1)+
            scale_y_continuous(expand = c(0,0),limits = c(0,a))+
            geom_hline(aes(yintercept=mean(mean)), colour="black", linetype=2) +
            geom_vline(aes(xintercept=0), colour="black", linetype="dashed") +
            labs(x=paste(name_i,"of all group", sep = "_"),
                 y="group",
                 title = paste("Normality test",p1,"Homogeneity of variance",p2,"kruskal.test",sumkrusk[3],sep = ":"))
          p
          p=p + Mytheme
          p
          if (length(unique(data_wt$group))>3){	p=p+theme(axis.text.x=element_text(angle=45,vjust=1, hjust=1))}
          FileName <- paste(plotname ,name_i,"_kruskal.test_nosig_bar", ".pdf", sep = "")
          ggsave(FileName, p, width = 8, height = 8)
        }


      }



    }
    if( plot == "box") {
      name_i = colnames(data_wt[i])

      if (p1 >.05& p2 >.05) {

        #进行方差检验 下面wtx3为提取的p值
        model<-aov(count ~ group, data= ss)#方差分析
        wtx1 = summary(model)
        wtx2 = wtx1[[1]]
        wtx3 = wtx2[5]#
        #p1 >=.05& p2 >=.05:数据符合正态分布，方差齐心
        if ( wtx3$`Pr(>F)`[1]< 0.05) {

          # data_box = data_wt[c(1,2,i)]
          # colnames(data_box) = c("ID" , "group","dd" )
          # out = LSD.test(model,"group", p.adj="none") # alternative fdr
          # stat = out$groups
          # stat

          if (method_Mc == "Tukey") {
            library(multcomp)

            litter.mc <- glht(model, linfct = mcp(group = 'Tukey'))
            # summary(litter.mc)
            insx = cld(litter.mc)
            aa <- insx$mcletters$monospacedLetters
            aa = as.data.frame(aa)
            colnames(aa) = c("groups")
            head(aa)
            aa$group = row.names(aa)
            stat = aa
          }


          if (method_Mc == "LSD") {
            out <- LSD.test(model,"group", p.adj="none")#进行多重比较，不矫正P值
            aa = out$group#结果显示：标记字母法
            aa$group = row.names(aa)
            aa
            stat = aa
          }

          #【SNK法】（Student-Newman-Keuls）程序运行结果与LSD.test类似。
          if (method_Mc == "SNK") {
            library("agricolae")
            out <- SNK.test(model,"group")
            aa = out$groups# 查看每个组的label

            aa$group = row.names(aa)
            stat = aa
            aa
          }

          #【Duncan法】(新复极差法)（SSR）
          if (method_Mc == "Duncan") {
            # library("agricolae")
            out <-duncan.test (model,"group")
            aa = out$groups# 查看每个组的label

            aa$group = row.names(aa)
            stat = aa
            aa
          }

          #Scheffe检验特点：各组样本数相等或不等均可以，但是以各组样本数不相等使用较多；
          if (method_Mc == "scheffe") {
            # library("agricolae")
            out <-scheffe.test (model,"group")
            aa = out$groups# 查看每个组的label

            aa$group = row.names(aa)
            stat = aa
            aa
          }


          data_box$stat=stat[as.character(data_box$group),]$groups
          max=max(data_box[,c("dd")])
          min=min(data_box[,c("dd")])
          x = data_box[,c("group","dd")]
          library(tidyverse)
          y = x %>% group_by(group) %>% summarise_(Max=paste('max(',"dd",')',sep=""))
          y=as.data.frame(y)
          y
          rownames(y)=y$group
          data_box$y=y[as.character(data_box$group),]$Max + (max-min)*0.05

          # mi=c("#1B9E77" ,"#D95F02", "#7570B3","#E7298A")
          FileName <- paste(plotname ,name_i,method_Mc,"aov_box", ".csv", sep = "_")
          write.csv(data_box,FileName,quote = F)


          p = ggplot(data_box, aes(x=group, y=data_box[["dd"]], color=group)) +
            geom_boxplot(alpha=1, outlier.size=0, size=0.7, width=0.5, fill="transparent") +
            labs(x=paste(name_i," group", sep = "_"),
                 y="group",
                 title = paste("Normality test",p1,"Homogeneity of variance",p2,sep = ":"))+
            geom_jitter( position=position_jitter(0.17), size=1, alpha=0.7)+theme(legend.position="none")+
            geom_hline(aes(yintercept=mean(data_box$dd)), colour="black", linetype=2) +
            geom_vline(aes(xintercept=0), colour="black", linetype="dashed")
          p
          if (sig_show == "abc") {
            p = p +
              geom_text(data=data_box, aes(x=group, y=y, color=group, label= stat))

            p
          }
          wtq = levels(data_wt$group)
          lis = combn(levels(data_wt$group), 2)
          x <-lis
          my_comparisons <- tapply(x,rep(1:ncol(x),each=nrow(x)),function(i)i)

          # if (sig_show == "line") {
          #
          #
          #
          #   p = p +
          #     stat_compare_means()+
          #     stat_compare_means(comparisons=my_comparisons,label = "p.signif",hide.ns = F) # Add pairwise
          #
          #   p
          # }

          if (sig_show == "line") {
            zuhe = combn(aa$group,2)
            xxxx <- tapply(zuhe,rep(1:ncol(zuhe),each=nrow(zuhe)),function(i)i)
            xxxx
            sig_lis = rep("a",dim(zuhe)[2])
            for (i in 1:dim(zuhe)[2]) {
              library(tidyverse)

              if (filter(aa, group == xxxx[[i]][1])$groups == filter(aa, group == xxxx[[i]][2])$groups) {
                sig_lis[i] = "no_sig"
              }

              if (filter(aa, group == xxxx[[i]][1])$groups != filter(aa, group == xxxx[[i]][2])$groups) {
                sig_lis[i] = "*"
              }

            }



            p = p +
              geom_signif(comparisons = xxxx, annotations=sig_lis,
                          y_position = (seq(from=1, to=max(data_box$dd)/4,length.out=dim(zuhe)[2]) + max(data_box$dd)), tip_length = rep(0.03,dim(zuhe)[2]),color = "black")
            p
          }

          p=p+Mytheme
          p
          if (length(unique(data_wt$group))>3){	p=p+theme(axis.text.x=element_text(angle=45,vjust=1, hjust=1))}
          FileName <- paste(plotname ,name_i,method_Mc,"aov_box", ".pdf", sep = "_")
          ggsave(FileName, p, width = 8, height = 8)

          #wtx3$`Pr(>F)`[1]>= 0.05当p值大于0.05那么就不能做多重比较了
        }else if ( wtx3$`Pr(>F)`[1]>= 0.05)  {

          data_box = data_wt[c(1,2,i)]
          colnames(data_box) = c("ID" , "group","dd" )
          max=max(data_box[,c("dd")])
          min=min(data_box[,c("dd")])
          x = data_box[,c("group","dd")]
          y = x %>% group_by(group) %>% summarise_(Max=paste('max(',"dd",')',sep=""))
          y=as.data.frame(y)
          rownames(y)=y$group
          data_box$y=y[as.character(data_box$group),]$Max + (max-min)*0.05
          FileName <- paste(plotname ,name_i,method_Mc,"_aov_nosig_box", ".csv", sep = "_")
          write.csv(data_box,FileName,quote = F)
          p = ggplot(data_box, aes(x=group, y=data_box[["dd"]], color=group)) +
            geom_boxplot(alpha=1, outlier.size=0, size=0.7, width=0.5, fill="transparent") +
            geom_hline(aes(yintercept=mean(as.vector(as.matrix(data_wt[i])))), colour="black", linetype=2) +
            geom_vline(aes(xintercept=0), colour="black", linetype="dashed") +
            labs(x=paste(name_i,"box", sep = "_"),
                 y="group",
                 title = paste("Normality test",p1,"Homogeneity of variance",p2,"aov",round(wtx3$`Pr(>F)`[1],3),sep = ":"))+
            geom_jitter( position=position_jitter(0.17), size=1, alpha=0.7)+theme(legend.position="none")
          p
          p=p+Mytheme
          p
          if (length(unique(data_wt$group))>3){	p=p+theme(axis.text.x=element_text(angle=45,vjust=1, hjust=1))}
          FileName <- paste(plotname ,name_i,"_aov_nosig_box", ".pdf", sep = "")
          ggsave(FileName, p, width = 8, height = 8)
        }
        # p1 = 1
        # p2 = 1


        #   p1 <.05| p2 <.05：不符合正态检验或者方差不齐性
      }else if (p1 <.05| p2 <.05){

        krusk=compare_means(count ~ group, data=ss, method = "kruskal.test")
        sumkrusk=as.data.frame(krusk)
        sumkrusk
        #多组比较小于0.05，表明多组之间具有差异，可以进行两两非参数检验，并标记字母

        if ( sumkrusk[3]< 0.05) {
          library(ggpubr)
          data_box = data_wt[c(1,2,i)]
          colnames(data_box) = c("ID" , "group","dd" )

          krusk=compare_means(count ~ group, data=ss, method = "wilcox.test")
          xx=as.data.frame(krusk)
          xx$group1
          wilcox_levels = paste(xx$group1,xx$group2,sep = "-")
          wilcox_levels = xx$p


          names(wilcox_levels) =  paste(xx$group1,xx$group2,sep = "-")
          library("multcompView")
          wilcox.labels <- data.frame(multcompLetters(wilcox_levels, threshold = 0.05)['Letters'])
          colnames(wilcox.labels) = "groups"
          aa = wilcox.labels
          aa = as.data.frame(aa)
          aa$group = row.names(aa)
          stat = aa

          data_box = data_wt[c(1,2,i)]
          colnames(data_box) = c("ID" , "group","dd" )


          data_box$stat=stat[as.character(data_box$group),]$groups
          max=max(data_box[,c("dd")])
          min=min(data_box[,c("dd")])
          x = data_box[,c("group","dd")]
          library(tidyverse)
          y = x %>% group_by(group) %>% summarise_(Max=paste('max(',"dd",')',sep=""))
          y=as.data.frame(y)
          y
          rownames(y)=y$group
          data_box$y=y[as.character(data_box$group),]$Max + (max-min)*0.05

          p = ggplot(data_box, aes(x=group, y=data_box[["dd"]], color=group)) +
            geom_boxplot(alpha=1, outlier.size=0, size=0.7, width=0.5, fill="transparent") +
            labs(x=paste(name_i," group", sep = "_"),
                 y="group",
                 title = paste("Normality test",p1,"Homogeneity of variance",p2,sep = ":"))+
            geom_jitter( position=position_jitter(0.17), size=1, alpha=0.7)+theme(legend.position="none")+
            geom_hline(aes(yintercept=mean(data_box$dd)), colour="black", linetype=2) +
            geom_vline(aes(xintercept=0), colour="black", linetype="dashed")
          p
          # sig_show = "abc"
          if (sig_show == "abc") {
            p = p +
              geom_text(data=data_box, aes(x=group, y=y, color=group, label= stat))

            p
          }

          FileName <- paste(plotname ,name_i,"_kruskal.test_wlc_box_", ".csv", sep = "_")
          write.csv(data_box,FileName,quote = F)
          wtq = levels(data_wt$group)
          lis = combn(levels(data_wt$group), 2)
          x <-lis
          my_comparisons <- tapply(x,rep(1:ncol(x),each=nrow(x)),function(i)i)

          if (sig_show == "line") {
            p = p +
              stat_compare_means()+
              stat_compare_means(comparisons=my_comparisons,label = "p.signif",hide.ns = F) # Add pairwise

            p
          }


          p
          p=p+Mytheme
          p
          if (length(unique(data_wt$group))>3){	p=p+theme(axis.text.x=element_text(angle=45,vjust=1, hjust=1))}
          FileName <- paste(plotname ,name_i,"_kruskal.test_wlc_box_", ".pdf", sep = "")
          ggsave(FileName, p, width = 8, height = 8)
        }else if ( sumkrusk[3] >= 0.05)  {
          data_box = data_wt[c(1,2,i)]
          colnames(data_box) = c("ID" , "group","dd" )

          max=max(data_box[,c("dd")])
          min=min(data_box[,c("dd")])
          x = data_box[,c("group","dd")]
          y = x %>% group_by(group) %>% summarise_(Max=paste('max(',"dd",')',sep=""))
          y=as.data.frame(y)
          rownames(y)=y$group
          data_box$y=y[as.character(data_box$group),]$Max + (max-min)*0.05
          data_box
          FileName <- paste(plotname ,name_i,method_Mc,"_kruskal.test_nosig_box", ".csv", sep = "_")
          write.csv(data_box,FileName,quote = F)
          p = ggplot(data_box, aes(x=group, y=data_box[["dd"]], color=group)) +
            # geom_hline(aes(yintercept=mean(as.vector(as.matrix(data_wt[i])))), colour="black", linetype=2) +
            # geom_vline(aes(xintercept=0), colour="black", linetype="dashed") +
            geom_boxplot(alpha=1, outlier.size=0, size=0.7, width=0.5, fill="transparent") +
            labs(x=paste(name_i,"box", sep = "_"),
                 y="group",
                 title = paste("Normality test",p1,"Homogeneity of variance",p2,"kruskal.test",round(wtx3$`Pr(>F)`[1],3),sep = ":"))+
            geom_jitter( position=position_jitter(0.17), size=1, alpha=0.7)+theme(legend.position="none")+
            stat_compare_means()
          p
          p=p + Mytheme
          p
          if (length(unique(data_wt$group))>3){	p=p+theme(axis.text.x=element_text(angle=45,vjust=1, hjust=1))}
          FileName <- paste(plotname ,name_i,"_kruskal.test_nosig_box", ".pdf", sep = "")
          ggsave(FileName, p, width = 8, height = 8)
        }


      }

    }



  }
}

}