Question: R: Manhattan Plot Of Fst Values Instead Of -Log(P)
3
Abdel360 wrote:

Hi all,

I have a tab delimited file with the following columns:

``````CHR  SNP  bp  Fst
``````

where Fst is the fixation index. Does anyone know how to make a Manhattan plot of these Fst values in R? Many thanks!

R visualization fst • 11k views
modified 5.4 years ago by CB10 • written 8.1 years ago by Abdel360
6
John1.5k wrote:

I found the following blog useful, you need tweak the codes to fit F-value not -log10(p) value - http://gettinggeneticsdone.blogspot.com/2011/04/annotated-manhattan-plots-and-qq-plots.html

Edits:

If you would like to use lattice or ggplot, we can create similar plot:

``````# just an example
Fst <- rnorm(10000, 10, 5)
chr <- c(rep(1, 2500), rep(2, 2500), rep(3, 2500), rep(4, 2500))
BP <- c(1:2500, 1:2500, 1:2500, 1:2500)
mydf <- data.frame (Fst, chr, BP)
require(ggplot2)
qplot(BP, Fst, facets = . ~ chr, col = factor(chr), data = mydf)
``````

I like bw theme better

``````qplot(BP, Fst, facets = . ~ chr, col = factor(chr), data = mydf) + theme_bw()
``````

You need to reshape plot area or change arrangement of graph to make all facets in single line as Manhattan plot. The graph is pretty close but not exactly as in above blog code.

You just have to replace bp by BP and Fst by P and it should work.

3
Zev.Kronenberg11k wrote:

I have done the exact plot you want for FST.

here is where [R] becomes awesome:

``````dat<-read.csv("your.data.txt", header=FALSE, sep="\t")
plot(dat\$FST, col=as.factor(dat\$CHR))
``````

If you need the true coordinates ie BP I will send you a script I wrote to take that into account. Otherwise the code above will suffice.

Pretty much R gives the X values a number `1:length(dat\$FST)` and then colors the points by the `dat\$CHR`

Hi, I tried this code for data in the same format as above, am I missing something or should these two R commands be sufficient for plotting? Error in plot.window(...) : need finite 'xlim' values. Thanks

please post `head(dat)`

Hi all,

I need to do the same plot as described above but I am facing the same problem.

``````plot(dat\$FST, col=as.factor(dat\$CHR))
Error in plot.window(...) : need finite 'xlim' values
1: In min(x) : no non-missing arguments to min; returning Inf
2: In max(x) : no non-missing arguments to max; returning -Inf
3: In min(x) : no non-missing arguments to min; returning Inf
4: In max(x) : no non-missing arguments to max; returning -Inf
``````

what I have to do? what do you mean by `head(dat)`?

Would you mind posting the script that you wrote to take BP into account? Cheers! :)

https://github.com/jewmanchue/zk_tools/blob/master/CHR_POS_GENOMIC_POS

First of all thanks for your help. I just need to use this code only

``````plot(dat\$FST, col=as.factor(dat\$CHR))
``````

but it give me the prevoius error. I dont need the true coordinates.

0
CB10 wrote:

I want to create a Manhattan plot to plot Fst values of each SNP. I used John's first code to plot Fst values for 3 different chromosomes and it worked well. But the size of chromosomes are fixed, and I want to set the maximum size for each chromosome. Is it possible? The code that I used was:

``````Fst <- rnorm(48167,0.13, 0.2)
chr <- c(rep(1, 14552), rep(2, 33582), rep(10, 33))
BP <- c(1:14552, 1:33582, 1:33)
data <- data.frame (Fst, chr, BP) qplot(BP, Fst, facets = . ~ chr, col = factor(chr), data = data)
``````

I tried to use the package qqman to plot Fst values instead of P values and did not work.

Or I need to use the perl script that Zev.Kronenberg mentioned? Convert chrmosomes positions into genomic coordiantes?

With ggplot2 you can facet on the chromosome: something like `+facet_grid(.~CHR)`. You will have to play with the margins on the facets, but it does work. If you use my script it make the process easier as you don't have to facet or fiddle with the coordinates.