PCA plot color differentiation

Question

[tothepoint]

I followed the post related to PCA from vcf file followed by

library("SNPRelate")
vcf.fn<-"~/xxx/tmp.vcf"
snpgdsVCF2GDS(vcf.fn, "ccm.gds",  method="biallelic.only")
genofile <- openfn.gds("ccm.gds") 
ccm_pca<-snpgdsPCA(genofile)
plot(ccm_pca$eigenvect[,1],ccm_pca$eigenvect[,2] ,col=as.numeric(substr(ccm_pca$sample, 1,3) == 'CCM')+3, pch=2)

I got a plot but not able to conclude from it. There is no legend or color code differentiation between the breeds. My understanding to R is very limited and couldn't able to find a fix. May I get some help/tips to fix this problem. I will be grateful to you all.

[Kevin Blighe]

You are already attempting to set colours via this parameter:

col=as.numeric(substr(ccm_pca$sample, 1,3) == 'CCM')+3

Did you write this code (above) on your own?

What is the output of:

ccm_pca$sample
substr(ccm_pca$sample, 1,3)

[tothepoint]

Dear Kevin, I found that code on previously mentioned post.

The output for

ccm_pca$sample

[1] "01" "03" "04" "05" "06"
[6] "07" "10:01" "10:03" "10:04" "10:05" [11] "10:06" "10:20" "20" "2:01" "2:03" [16] "2:04" "2:05" "2:06" "2:07" "2:20" [21] "3:01" "3:03" "3:04" "3:05" "3:06" [26] "3:07" "3:20" "4:01" "4:03" "4:04" [31] "4:05" "4:06" "4:07" "4:20" "5:01" [36] "5:03" "5:04" "5:05" "5:06" "5:07"

substr(ccm_pca$sample, 1,3)

[1] "01" "03" "04" "05" "06" "07" "10:" [8] "10:" "10:" "10:" "10:" "10:" "20" "2:0" [15] "2:0" "2:0" "2:0" "2:0" "2:0" "2:2" "3:0" [22] "3:0" "3:0" "3:0" "3:0" "3:0" "3:2" "4:0" [29] "4:0" "4:0" "4:0" "4:0" "4:0" "4:2" "5:0" [36] "5:0" "5:0" "5:0" "5:0" "5:0

Respectively

[tothepoint]

Thanks for sharing..!!

Although this works perfectly I found one issue regarding grouping of samples manually. What if we have to color this group into forest green.

'03', '10':03', '2:03'

Similarly, this group into red color.

'05', '10:05', '3:05'

Any suggestion.

[Kevin Blighe]

Is the Reds or Greens palette sufficient?

Alternatively, you can literally use any colours that you want, like this:

myPalette <- c('white','pink','red2','red4')
colorRampPalette(myPalette)(length(unique(x)))[factor(x, levels = c('01', '03', '05', '08'))]

Answer 1

[Kevin Blighe]

I am confident that you know to what these numbers relate?

set specific colours for each factor level

You can set colours manually to any vector like this:

x <- c('01','03','05','01','08','08','01','03','05')
unique(x)
[1] "01" "03" "05" "08"
colours <- c('royalblue', 'forestgreen', 'red4', 'pink')[factor(x, levels = c('01', '03', '05', '08'))]

Here, the order of the colours ('royalblue', 'forestgreen', 'red4', 'pink') must match the order of the factor levels that you are setting ('01', '03', '05', '08').

apply a colour gradient / pre-built palette to your factor levels

You can also automatically assign colours via RColorBrewer:

require('RColorBrewer')
pick.col <- brewer.pal(10, 'Spectral')
colII <- colorRampPalette(pick.col)(length(unique(x)))[factor(x, levels = c('01', '03', '05', '08'))]

Here, I use the 'Spectral' palette. For other palettes, take a look here (half-way): http://www.sthda.com/english/wiki/colors-in-r

create your own palette

For a custom palette, do something like this:

myPalette <- c('white','pink','red2','red4')
colorRampPalette(myPalette)(length(unique(x)))[factor(x, levels = c('01', '03', '05', '08'))]

----------------------

You can then plot a custom legend via the legend() function, which comes with base R.

Kevin