0

I am trying to transform the code of a plot from base R to taking advantage of the plotly library. Here is the code for the plot, as well as an image. contrastTable is a table that is being read in (above the code here) that contains contrasts between groups for different GSE #'s. (see here for more info) 

#Get gene to plot
  gene = toupper(readline(prompt="Enter gene name (e.g. GZMA): "))
  #Find all instances of query gene in master contrast table
  geneLines = grep(paste("^",gene,"$",sep = ""),contrastTable[,"gene"])
  #If the gene is found at least once, continue
  if (length(geneLines)>0)
  {
    ########Plot contrasts
    par(oma = c(1,10,1,1))
    geneRows = contrastTable[geneLines,]
    fcs = as.list(geneRows$logFC)
    names(fcs) = paste(geneRows$GSE,"\n",geneRows$Group2," \U2192 ",geneRows$Group1,sep = "")
    maxFC = max(as.numeric(as.character(geneRows$logFC)))
    minFC = min(as.numeric(as.character(geneRows$logFC)))
    #Set up marker color and type
    markers = rep(4,length(fcs))
    markers[which(as.numeric(as.character(geneRows[,"adj.P.Val"]))<0.05)] = 15 
    markers[which(as.numeric(as.character(geneRows[,"adj.P.Val"]))<0.01)] = 19 
    outerCol = rep("grey20",length(fcs))
    outerCol[which(as.numeric(as.character(geneRows[,"adj.P.Val"]))<0.05)] = "darkorange" 
    outerCol[which(as.numeric(as.character(geneRows[,"adj.P.Val"]))<0.01)] = "red" 
    layout(t(c(1,2)), c(3.5,1), c(1,1))
    #Plot all contrasts
    stripchart(fcs,xlim = c(min(-2,minFC),max(2,maxFC)),las = 1,col = outerCol,pch = markers,main = gene,xlab = expression(paste("log"[2],"(Fold Change)")),cex.axis=0.75)
    abline(v = 0,col = "grey50",lty=2)
    abline(v = log(c(0.5,2/3,1.5,2),2),col = "grey80",lty=2)
    usr = par("usr")
    legend(usr[1]-(usr[2]-usr[1])/3.5,usr[3]-(usr[4]-usr[3])/10,c("p \u2265 0.05","p < 0.05","p < 0.01"),col = c("black","darkorange","red"),pch = c(4,15,19),cex = 0.7,xpd = TRUE)
    #add confidence interval
    for (i in 1:nrow(geneRows))
    {
      segments(geneRows[i,"CI.L"],i,geneRows[i,"CI.R"],i,col = "grey50")
    }
    #Re-plot to put markers on top
    par(new=TRUE)
    stripchart(fcs,xlim = c(min(-2,minFC),max(2,maxFC)),las = 1,col = outerCol,pch = markers,xlab = "",ylab = "",xaxt = "n",yaxt="n")
    #add lines dividing different GSE#s
    datasetList = geneRows$GSE
    hLines =(which(duplicated(datasetList) == FALSE) - 0.5)
    abline(h = hLines, col = "grey80")

plot

Aside from the obvious spacing/scaling issues with the plot, I have two main questions.

1) How can I recreate the horizontal lines between each GSE # in plotly? (it is now currently being done by the last two lines of code)

2) How can I recreate the y-axis in plotly, where it shows the GSE #, and underneath, the group 1 -> group 2 kind of format?

Any help with this would be greatly appreciated!

Kyle Weise
  • 869
  • 1
  • 8
  • 29
  • 1) https://stackoverflow.com/questions/34093169/horizontal-vertical-line-in-plotly – S Rivero Jul 07 '17 at 13:19
  • 2) Maybe this helps https://stackoverflow.com/questions/38260022/line-break-and-subscript-in-axis-title-using-plotly-in-r – S Rivero Jul 07 '17 at 13:22
  • @SRivero That second link helps me out with something else actually..but the problem with the first one link is that I can't hard code the x/y points (as is shown) for the lines because they are variable (i.e. there is a different # of GSE #'s shown based on what the gene selected is). – Kyle Weise Jul 07 '17 at 13:32

0 Answers0