Is it possible to group multiple bigWig files in Gviz DataTrack?
1
0
Entering edit mode
@romanhillje-13023
Last seen 15 months ago

Hi everybody,

I was wondering if it was possible to group different DataTrack objects in a Gviz plot? In the user guide, the example for grouping data is done using a GRanges objects that contains two metadata columns for the different scores that belong to a group. Creating this is difficult because I would like to group scores/coverages from different bigWig files that have different intervals and therefore can't simply be merged together into a single GRanges object. I already started exploring possible strategies to do that, but it seems quite complicated, especially since I'm planning to plot around 100 profiles, which is quite easy if they don't need to be grouped, but with grouping the plot would be a lot nicer. You can find the corresponding question on Biostars here: https://www.biostars.org/p/448502/

My current way to plot the tracks follows the user guide, reading the bigWig files into separate DataTrack objects, then creating a list from them and handing that list over to plotTrack (see below).

library(Gviz)

gtrack <- GenomeAxisTrack()
itrack <- IdeogramTrack(genome = 'mm10', chromosome = 'chr5')

txdb <- GenomicFeatures::makeTxDbFromGFF('annotation.gtf.gz')

grtrack <- GeneRegionTrack(
  txdb,
  genome = 'mm10',
  chromosome = 'chr5',
  name = 'RefSeq'
)

bigwig_files <- paste0(
  '<input_dir>/',
  rep(c('H3K4me3','H3K27me3','H3K27ac','H3K4me1'), each = 4),
  '_',
  c('SD_3m','SD_6m','SD_12m','CR_12m'),
  '.bw'
)

track_names <- gsub(bigwig_files, pattern = '.bw', replacement = '')

counter <- 0
for ( file in bigwig_files ) {
  counter <- counter + 1
  if ( grepl(bigwig_files[counter], pattern = 'H3K4me3') ) {
    gradient <- RColorBrewer::brewer.pal(9, 'Blues')
  } else if ( grepl(bigwig_files[counter], pattern = 'H3K27me3') ) {
    gradient <- RColorBrewer::brewer.pal(9, 'Reds')
  } else if ( grepl(bigwig_files[counter], pattern = 'H3K27ac') ) {
    gradient <- RColorBrewer::brewer.pal(9, 'Greens')
  } else if ( grepl(bigwig_files[counter], pattern = 'H3K4me1') ) {
    gradient <- RColorBrewer::brewer.pal(9, 'Oranges')
  }
  assign(
    x = track_names[counter],
    value = DataTrack(
      range = bigwig_files[counter],
      genome = 'mm10',
      type = 'heatmap',
      name = track_names[counter],
      size = 2,
      gradient = gradient,
      window = -1,
      chromosome = 'chr5'
    )
  )
}

my_list <- list(itrack, gtrack, grtrack)
for ( i in 1:length(bigwig_files) ) {
  my_list[[i+3]] <- get(track_names[i])
}

plotTracks(
  my_list,
  from = 142903115-10000,
  to = 142906754+10000
)

And this is the resulting plot.

Gviz plot

I know that those are heatmaps, but if I can manage to group them together I would like to use lines, similar to this example from the user guide (chapter 4.3.1, first figure, https://bioconductor.org/packages/release/bioc/vignettes/Gviz/inst/doc/Gviz.html#43_datatrack).

I also wasn't able to make the track names fit into the little boxes, which could've been a possible, yet not idea, workaround.

I would greatly appreciate if somebody could give me some advise if what I have in mind is even possible, and if so, how to achieve it. Or, if it's not possible, what some possible workarounds could be.

Best, Roman

Gviz • 454 views
ADD COMMENT
0
Entering edit mode
Robert Ivanek ▴ 700
@robert-ivanek-5892
Last seen 4 days ago
Switzerland

Hi Roman, You can use OverlayTrack to overlay those 4 samples as a line plot in one graph. You would need to always tweak the y-limits of first sample in each group to cover the data from overlapped samples.

The solution proposed on biostars forum might also work but it is not ideal. How would you treat cases if the "tile" overlaps with two regions in GRanges (derived from WIG file)?

Best Robert

BTW. In my opinion cross-posting is really bad habit.

ADD COMMENT
0
Entering edit mode

Ok, that would still be a workaround however, right? Since I won't be able to combine the data into mean + confidence interval as shown in the figure from the user guide that I was referring to. And even if it's just the separate lines, it doesn't really help that much. The problem is that the track labels are not shown. I can't manually add them later, that's too much work. If the tracks in the plot I showed would be grouped, I could have a group label "H3K4me3" and then the individual track names as sample names.

For the exact reason you mentioned I was doubtful about the solution proposed on Biostars (actually, that's why I came here). But I don't understand your comment. Would you have noticed my question there? Should I have deleted my question on Biostars? They are not even identical questions. Here I specifically ask about Gviz, whereas on Biostars it's about merging GRanges objects.

Best, Roman

ADD REPLY
0
Entering edit mode

But in that case the best would be to pre-calculated manually, to obtain GRanges object with individual score columns. If the bigWig files do not have fixed with it might be necessary to have 1bp tiles to avoid above mentioned case that one tile overlaps with multiple ranges in one bigWig file.

ADD REPLY

Login before adding your answer.

Traffic: 327 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6