How to use pophelper to create “smoother” ADMIXTURE plots

A couple of years ago, when I started out with bioinformatics, I had to create ADMIXTURE plots. After generating the ancestry fractions using ADMIXTURE, I used pophelper to create them. This is what my plot looked like (I recreated it using the 1000 Genomes Projects dataset):

And this is what I wanted my plot to look like:

I wanted it to look smoother.

The problem I faced was that I could not reorder the population groups (e.g. LWK, ESN) using subsetgrp and sort individuals by their ancestry proportions using sortind="all" simultaneously. Here is the solution¹.

Reading the files and creating population labels

I created a data frame called popnames where column 1 (V1) contains the population names listed for each sample in order, and column 2 (V2) contains the respective continental populations.

I set the stringsAsFactors as FALSE so that the population labels can be read by grplab.

In this example, I am using a K value of 8.

library(pophelper)

# population names
popnames <- read.table("pop.txt", stringsAsFactors = FALSE)

# K=8
alist8 <- readQ(files="ADMIX.8.Q")

Reordering the population groups

The next step was to reorder the populations using subsetgrp. However, instead of creating a plot, the data, which will contain the reordered population groups, will be saved to a variable.

• grplab — assigns the population labels
• selgrp = "V1" — because there are two population labels, I have to declare the one I’ll be using for ordergrp and subsetgrp
• ordergrp = TRUE — in order to sort them using subsetgrp, I need to set this as TRUE so that all individuals from the same population are grouped together
• returndata = TRUE — instead of plotting the data, the data will be stored as a variable
• exportplot = FALSE — does not create a plot

neworder <- plotQ(alist8, grplab=popnames[,c(1,2)],
                  selgrp = "V1",
                  ordergrp = TRUE,
                  subsetgrp = c("LWK", "ESN", "YRI", "MSL", "GWD", "ACB", "ASW", "TSI", "IBS", "GBR", "CEU", "FIN", "GIH", "PJL", "BEB", "STU", "ITU", "KHV", "CDX", "CHS", "CHB","JPT", "PUR", "CLM", "MXL", "PEL"),
                  returndata = TRUE,
                  exportplot = FALSE)

Creating an ADMIXTURE plot

Next, I created the plot. The data is located in neworder$data$qlist and the population labels are located in neworder$data$grplab[[1]].

• sortind = "all" — all clusters (i.e. ancestry groups) are considered when sorting out the individuals
• exportpath=getwd() — export the plot in the current working directory

qdata2 <- plotQ(neworder$data$qlist, grplab=neworder$data$grplab[[1]],
                ordergrp = FALSE,
                selgrp = "V1",
                sortind = "all",
                height = 5,
                grplabheight = 1,
                showdiv = TRUE, divsize = 0.4,
                clustercol = c("#1D72F5","#DF0101","#77CE61", "#FF9326","#A945FF","#0089B2","#FDF060","#FFA6B2"),
                exportpath=getwd())

And voilà! A smoother, more satisfying plot.

1. I believe the author of pophelper posted about this, but I can’t seem to find it. If I do, I will mention it on this blog post. ↩