R/plot_con_effectsize.R

In LyonsZhang/ProgPerm: Progressive permutation for a dynamic representation of the robustness of microbiome discoveries

plot_con_effectsize<-function(bestoutputs,variable,testdata,estitle=NULL,saveesfile,espicdim,crtitle=NULL,savecrfile,crpicdim,goodtype="goodpvname",effecttext=FALSE)
{
  select<-switch(goodtype,
         goodpsigname=bestoutputs$goodpsigname,
         goodpvname=bestoutputs$goodpvname,
         goodesname=bestoutputs$goodesname,
         goodcrname=bestoutputs$goodcrname,
         stop("goodtype must be one of the follows: 'goodpsigname', 'goodpvname', 'goodesname' and 'goodcrname'. "))
  selname<-bestoutputs$selname

  #######plot effectsize
  df<-data.frame(matrix(0,length(selname),2))
  row.names(df)<-selname
  colnames(df)<-c("selname","effectsize")
  df$selname<-selname
  df$effectsize<-bestoutputs$effectsize0
  df1<-df[selname %in% select,]
  df1$signed<-sign(df1$effectsize)
  temp<-df1[df1$signed==-1,]
  temp<- temp[order(temp$effectsize),]
  df1<-rbind(df1[df1$signed==1,],temp)
  df1$selname<-factor(df1$selname,levels=df1$selname)
  df1[[variable]]<-factor(ifelse(df1$signed==1,'positive','negative'))

  p<-ggplot(df1, aes(selname,effectsize)) +
    geom_bar(stat = "identity", aes_string(fill = variable)) +
    scale_fill_manual(values=c("#00b0f6","#f8766d"))+
    geom_hline(yintercept=0,size=2)+
    ylab("Effect size") +
    xlab("Selected features with decreasing order")+
    coord_flip()+
    ggtitle(estitle) +
    theme_bw()+
    theme(plot.title = element_text(size = 30, face = "bold"),
          legend.title=element_text(size=20),
          legend.text=element_text(size=16))+
    theme(
      #axis.ticks.length =unit(10,"mm"),
      #axis.ticks.y = element_line(size=0.5),
      axis.line = element_line(colour = "black"),
      #panel.grid.major = element_blank(),
      panel.border = element_blank(),
      panel.background = element_blank(),
      axis.line.x = element_line(color="black", size = 1),
      axis.line.y = element_line(color="black", size = 1),
      axis.text=element_text(size=16,face="bold"),
      axis.title=element_text(size=20,face="bold"))
  if(effecttext==TRUE)
  {
    p<-p+geom_text(aes(label = round(effectsize,2),
                  hjust=0.5,
                  angle = 0))
  }
  p
  ggsave(filename=saveesfile,width = espicdim[1], height = espicdim[2])

  #######plot correlations
  df<-data.frame(matrix(0,length(selname),2))
  row.names(df)<-selname
  colnames(df)<-c("selname","correlation")
  df$selname<-selname
  df$correlation<-bestoutputs$correlation0
  df2<-df[selname %in% select,]
  df2$signed<-sign(df2$correlation)
  temp<-df2[df2$signed==-1,]
  temp<- temp[order(temp$correlation),]
  df2<-rbind(df2[df2$signed==1,],temp)
  df2$selname<-factor(df2$selname,levels=df2$selname)
  df2[[variable]]<-factor(ifelse(df2$signed==1,'positive','negative'))


  p<-ggplot(df2, aes(selname,correlation)) +
    geom_bar(stat = "identity", aes_string(fill = variable)) +
    scale_fill_manual(values=c("#00b0f6","#f8766d"))+
    geom_hline(yintercept=0,size=2)+
    ylab("Correlation") +
    xlab("Selected features with decreasing order")+
    coord_flip()+
    ggtitle(crtitle) +
    theme_bw()+
    theme(plot.title = element_text(size = 30, face = "bold"),
          legend.title=element_text(size=20),
          legend.text=element_text(size=16))+

    theme(
      #axis.ticks.length =unit(10,"mm"),
      #axis.ticks.y = element_line(size=0.5),
      axis.line = element_line(colour = "black"),
      #panel.grid.major = element_blank(),
      panel.border = element_blank(),
      panel.background = element_blank(),
      axis.line.x = element_line(color="black", size = 1),
      axis.line.y = element_line(color="black", size = 1),
      axis.text=element_text(size=16,face="bold"),
      axis.title=element_text(size=20,face="bold"))
  if(effecttext==TRUE)
  {
    p<-p+geom_text(aes(label = round(correlation,2),
                       hjust=0.5,
                       angle = 0))
  }
  p
  ggsave(filename=savecrfile,width = crpicdim[1], height = crpicdim[2])

  return(list(sigeffectsize=df1,sigcorrelation=df2))
}