R script for 3d PCOA plotting?

Integrating mothur with other programs
1

Hi all,

I got the pcoa results from mothur, named “xx.pcoa.axes”. I would like to plot the results as the 3d pcoa plot. Thanks to the script provided by dwaite, I was able to plot 2d pcoa. The code as below, in case it might be useful to someone else

pcoa_data = read.table(file="10_horsepick.thetayc.0.03.lt.pcoa.axes", header=T, sep='\t')
design_file = read.table(file="10_horse.design", sep='\t',stringsAsFactors=FALSE)
uniq_groups = unique(design_file[,2])
num_uniq_groups = length(uniq_groups)

avail_cols = rainbow(num_uniq_groups)
avail_symbols = 1:num_uniq_groups

colour_map = list()
symbol_map = list()

for(i in 1:num_uniq_groups) {
  current_group = uniq_groups[i]
  colour_map[[ current_group ]] = avail_cols[i]
  symbol_map[[ current_group ]] = avail_symbols[i]
}

col_vector = rep(NA,nrow(design_file))
symbol_vector = rep(NA,nrow(design_file))

for(i in 1:nrow(design_file)) {
  group = design_file[i,2]
  col_vector[i] = colour_map[[ group ]]
  symbol_vector [i] = symbol_map [[ group ]]   
}

plot(pcoa_data$axis2~pcoa_data$axis1,xlab='Axis 1',ylab='Axis 2',main='PCoA',col='black', pch=symbol_vector)
legend(X,Y,uniq_groups,avail_cols)

It would be nice to assign different colors to each group as well.

Thanks many!!

2

For 3D plotting I’m quite a fan of the rgl package in R. It’s very simple to get started with, basically just:

library(rgl)
plot3d(x=..., y=..., z=...)

As well as the typical static image exports, you can also export your graphs into an HTML file using the writeWebGL function which creates a graph that anyone can open in their web browser and rotate/zoom as required.

For what it’s worth, although I wrote that script a long time ago I’m not actually a big fan of it. I usually find it much better to manually choose and apply colours to the plot rather than rely on the automatically generated ones.