Stack several manhattan plots

How can I stack several manhattan plots in the same plot looks like this https://imgur.com/a/pj40c using R?

That plot is made using excel and not follow their real map position.

In my case I have 4 p-values from 4 different phenotypes. They are estimated from same SNPs. I want to stack them in the same plot but distinguish phenotypes by different colors.

   # Input data format, example. 
   SNP  CHR POS pval_1  pval_2  pval_3  pval_4
    a   1   100 0.1 0.5 0.2 0.1
    b   2   110 0.2 0.6 0.3 0.5
    c   3   120 0.3 0.7 0.1 0.6
    d   4   130 0.4 0.1 0.4 0.2

qqman is a very useful R package, unfortunately it is only allowed for single trait.

manhattan qqman r

• 12,294 views • link

updated 3.5 years ago by nayer •

• written 8.8 years ago by SheelS •

3 answers

With a bit of customization you could do it in R with ggplot:

library(dplyr)
library(ggplot2)

tmp <- data_frame(
  POS=seq(1,10000),
  pval1=1/runif(n = 10000, min=0, max=0.2),
  pval2=1/c(runif(n = 9999, min=0, max=0.2), 1e-5),
  pval3=1/runif(n = 10000, min=0, max=0.3),
  pval4=1/runif(n = 10000, min=0, max=0.4))

tmp.tidy <- tmp %>% gather(key, value, -POS)

ggplot(tmp.tidy, aes(POS, value, color=key)) + 
  geom_point()

Example:

https://ibb.co/mYnb3w

Use the + facet_wrap(~CHR) to create a graph per chromosome.

• 0 views • link

written 8.8 years ago by VHahaut

Thanks, I thought the same thing before posting my question. But I wonder that do you know how to create a column for CHR using my example data (for + facet_wrap(~CHR) usage ) ? Thank you.

• 0 views • link

written 8.8 years ago by SheelS •

Using the facet_wrap you'll split your graph in as many chromosome as you have:

library(dplyr)
library(ggplot2)

tmp <- data_frame(
  CHR=paste0("chr", sample(1:10, 1000, replace=T)),
  POS=seq(1,1000),
  pval1=1/runif(n = 1000, min=0, max=0.2),
  pval2=1/c(runif(n = 999, min=0, max=0.2), 1e-5),
  pval3=1/runif(n = 1000, min=0, max=0.3),
  pval4=1/runif(n = 1000, min=0, max=0.4))

tmp.tidy <- tmp %>% gather(key, value, -POS, -CHR)

ggplot(tmp.tidy, aes(POS, value, color=key)) + 
  geom_point() + facet_wrap(~CHR, nrow=1)

With the facet_wrap() by chromosome:

http://hpics.li/3db6911

• 0 views • link

written 8.8 years ago by VHahaut

You can now do it with karyoploteR.

In the karyoploteR tutorial there's a specific page on how to create manhattan plots and it has different plots combining multiple manhattans in a single plot, but none of them is by overlaying them as you ask. To do that you would simply need to call kpPlotManhattan multiple times with different datasets and that would be it.

kp <- plotKaryotype(plot.type=4)
kpPlotManhattan(kp, data1, points.col="red")
kpPlotManhattan(kp, data2, points.col="blue")
kpPlotManhattan(kp, data3, points.col="black")

These are other example plots combining multiple manhattan plots.

enter image description here