Colour SNP points according to annotation2 rather than chromosome?

[sophiegresh]

Hi Sahir,

First of all thanks for creating such a fantastic package! I've been trying and failing to edit some of the code in the package for a while now in order to recolour the SNP points according to another variable rather than chromosome number in the manhattan plots, and I was wondering whether you could help? I want to colour the SNPs according to geographical region (of which I have two) as I have multiple populations plotted on the same manhattan plot. When transforming my dataset into the manhattanr class I saved this region variable as the annotation2 so each SNP has either the region "Africa" or "MiddleEast" in the annotation2 column. So is there a way I can colour points according to annotation2 rather than chromosome?

Many Thanks,
Sophie

[sahirbhatnagar]

Thanks for your interest in this package. Several people have asked for this feature. I will try to implement it in a function.

for the time being, here is some code that should get you started:

library(manhattanly)
library(magrittr)

DT <- subset(HapMap, CHR %in% 1:3)
DT$region <- sample(c("Africa", "MiddleEast"), size = nrow(DT), replace = TRUE)

x <- manhattanr(DT, snp = "SNP", gene = "GENE", annotation1 = "region")

d <- x$data
pName <- x$pName
snpName <- x$snpName
geneName <- x$geneName
annotation1Name <- x$annotation1Name
annotation2Name <- x$annotation2Name
labs <- x$labs
xlabel <- x$xlabel
ticks <- x$ticks
nchr <- x$nchr
xmax = ceiling(max(d$pos) * 1.03)
xmin = floor(max(d$pos) * -0.03)

p <- plotly::plot_ly()
p %<>% plotly::layout(p, title = "Title", 
                        xaxis = list(title = "Chromosome", 
                        showgrid = FALSE, range = c(xmin, xmax), autotick = FALSE, 
                        tickmode = "array", tickvals = ticks, ticktext = labs, 
                        ticks = "outside"), yaxis = list(title = "-log10(p)"))

icol <- 1
for (i in unique(d$index)) {
  tmp <- d[d$index == unique(d$index)[i], ]
  TEXT <- paste(if (!is.na(snpName)) 
    paste0(snpName, ": ", tmp[[snpName]]), if (!is.na(geneName)) 
      paste0(geneName, ": ", tmp[[geneName]]), if (!is.na(annotation1Name)) 
        paste0(annotation1Name, ": ", tmp[[annotation1Name]]), 
    if (!is.na(annotation2Name)) 
      paste0(annotation2Name, ": ", tmp[[annotation2Name]]), 
    sep = "<br>")
  
  chromo <- unique(tmp[which(tmp$index == i), "CHR"])
  
  p %<>% add_markers(x = tmp$pos, y = tmp$logp, color = tmp$region, 
                       showlegend = F, text = TEXT, name = paste0("chr", chromo))
  
  icol = icol + 1
}
p

See the resulting plot here:

https://plot.ly/~sahirbhatnagar/146/title/

[sophiegresh]

Thanks so much Sahir, I've managed to integrate that into the function code that I have and it's working beautifully!

[ghost]

Hi Sahir,

Thanks for creating such a useful package. On a similar note, is it possible to highlight snps/genes in more than just one color? It seems like the current function only accepts a single color. It would be great if the function could be extended to highlight1, highlight2, as you did with annotate.

Thanks again for the contribution!

[vdejager]

Hi Sahir,
I would vote for highlighting in multiple colors as well