Introduction

There are two packages available in Bioconductor for visualizing genomic data: rtracklayer and Gviz. rtracklayer provides an interface to genome browsers and associated annotation tracks. Gviz can be used to plot coverage and annotation tracks. TrackViewer is a lightweight visualization tool for generating interactive figures for publication. Not only can trackViewer be used to visualize coverage and annotation tracks, but it can also be employed to generate lollipop/dandelion plots that depict dense methylation/mutation/variant data to facilitate an integrative analysis of these multi-omics data. It leverages Gviz and rtracklayer, is easy to use, and has a low memory and cpu consumption. In addition, we implemented a web application of trackViewer leveraging Shiny package. The web application of trackViewer is available at https://github.com/jianhong/trackViewer.documentation/tree/master/trackViewerShinyApp.

library(Gviz)
library(rtracklayer)
library(trackViewer)
extdata <- system.file("extdata", package="trackViewer",
                       mustWork=TRUE)
gr <- GRanges("chr11", IRanges(122929275, 122930122), strand="-")
fox2 <- importScore(file.path(extdata, "fox2.bed"), format="BED",
                    ranges=gr)
fox2$dat <- coverageGR(fox2$dat)

viewTracks(trackList(fox2), gr=gr, autoOptimizeStyle=TRUE, newpage=FALSE)

dt <- DataTrack(range=fox2$dat[strand(fox2$dat)=="-"] , 
                genome="hg19", type="hist", name="fox2", 
                window=-1, chromosome="chr11", 
                fill.histogram="black", col.histogram="NA",
                background.title="white",
                col.frame="white", col.axis="black",
                col="black", col.title="black")
plotTracks(dt, from=122929275, to=122930122, strand="-")
Plot data with **Gviz** and **trackViewer**. Please note that **trackViewer** can generate similar figure as **Gviz** with several lines of simple codes.

Plot data with Gviz and trackViewer. Please note that trackViewer can generate similar figure as Gviz with several lines of simple codes.

trackViewer not only has the functionalities to produce the figures generated by Gviz, as shown in the Figure above, but also provides additional plotting styles as shown in the Figure below. The mimimalist design requires minimum input from the users while retaining the flexibility to change the output style easily.

gr <- GRanges("chr1", IRanges(c(1, 6, 10), c(3, 6, 12)), score=c(3, 4, 1))
dt <- DataTrack(range=gr, data="score", type="hist")
plotTracks(dt, from=2, to=11)
tr <- new("track", dat=gr, type="data", format="BED")
viewTracks(trackList(tr), chromosome="chr1", start=2, end=11)
Plot data with **Gviz** and **trackViewer**. Note that **trackViewer** is not only including more details but also showing all the data involved in the given range.

Plot data with Gviz and trackViewer. Note that trackViewer is not only including more details but also showing all the data involved in the given range.

Gviz requires huge memory space to handle big wig files. To solve this problem, we rewrote the import function in trackViewer by importing the entire file first and parsing it later when plot. As a result, trackViewer decreases the import time from 180 min to 21 min and the memory cost from 10G to 5.32G for a half giga wig file (GSM917672).

Browse

Steps of using trackViewer

Step 1. Import data

The function importScore is used to import BED, WIG, bedGraph or BigWig files. The function importBam is employed to import the bam files. Here is an example.

library(trackViewer)
extdata <- system.file("extdata", package="trackViewer",
                       mustWork=TRUE)
repA <- importScore(file.path(extdata, "cpsf160.repA_-.wig"),
                    file.path(extdata, "cpsf160.repA_+.wig"),
                    format="WIG")
## Because the wig file does not contain any strand info, 
## we need to set it manually.
strand(repA$dat) <- "-"
strand(repA$dat2) <- "+"

The function coverageGR could be used to calculate the coverage after the data is imported.

fox2 <- importScore(file.path(extdata, "fox2.bed"), format="BED",
                    ranges=GRanges("chr11", IRanges(122929000, 122931000)))
dat <- coverageGR(fox2$dat)
## We can split the data by strand into two different track channels
## Here, we set the dat2 slot to save the negative strand info. 
 
fox2$dat <- dat[strand(dat)=="+"]
fox2$dat2 <- dat[strand(dat)=="-"]

Step 2. Build the gene model

The gene model can be built for a given genomic range using geneModelFromTxdb function which uses the TranscriptDb object as the input.

library(TxDb.Hsapiens.UCSC.hg19.knownGene)
library(org.Hs.eg.db)
gr <- GRanges("chr11", IRanges(122929275, 122930122), strand="-")
trs <- geneModelFromTxdb(TxDb.Hsapiens.UCSC.hg19.knownGene,
                         org.Hs.eg.db,
                         gr=gr)

Users can generate a track object with the geneTrack function by inputting a TxDb and a list of gene Entrez IDs. Entrez IDs can be obtained from other types of gene IDs such as gene symbol by using the ID mapping function. For example, to generate a track object given gene FMR1 and human TxDb, refer to the code below.

entrezIDforFMR1 <- get("FMR1", org.Hs.egSYMBOL2EG)
theTrack <- geneTrack(entrezIDforFMR1,TxDb.Hsapiens.UCSC.hg19.knownGene)[[1]]

Step 3. View the tracks

Use viewTracks function to plot data and annotation information along genomic coordinates. addGuideLine or addArrowMark can be used to highlight a specific region.

viewerStyle <- trackViewerStyle()
setTrackViewerStyleParam(viewerStyle, "margin", c(.1, .05, .02, .02))
vp <- viewTracks(trackList(repA, fox2, trs), 
                 gr=gr, viewerStyle=viewerStyle, 
                 autoOptimizeStyle=TRUE)
addGuideLine(c(122929767, 122929969), vp=vp)
addArrowMark(list(x=122929650, 
                  y=2), # 2 means track 2 from the bottom.
             label="label",
             col="blue",
             vp=vp)
plot data and annotation information along genomic coordinates

plot data and annotation information along genomic coordinates

Adjust the styles

Adjust the x-axis or the X scale

In most cases, researchers are interested in the relative position of the peaks in the gene. Sometimes, margin needs to be adjusted to be able to show the entire gene model. The Figure below shows how to add an X scale (x-scale) and remove the x-axis using the setTrackXscaleParam and setTrackViewerStyleParam functions.

optSty <- optimizeStyle(trackList(repA, fox2, trs))
trackList <- optSty$tracks
viewerStyle <- optSty$style
setTrackViewerStyleParam(viewerStyle, "xaxis", FALSE)
setTrackViewerStyleParam(viewerStyle, "margin", c(.01, .05, .01, .01))
setTrackXscaleParam(trackList[[1]], "draw", TRUE)
setTrackXscaleParam(trackList[[1]], "gp", list(cex=0.8))
viewTracks(trackList, gr=gr, viewerStyle=viewerStyle)
plot data with x-scale

plot data with x-scale

setTrackXscaleParam(trackList[[1]], attr="position", 
                    value=list(x=122929700, y=3, label=200))
viewTracks(trackList, gr=gr, viewerStyle=viewerStyle)
plot data with x-scale

plot data with x-scale

Adjust the y-axis

The y-axis can be put to the right side of the track by setting the main slot to FALSE in the y-axis slot of each track. In addition, the limit of y-axis (ylim) can be set by setTrackStyleParam.

setTrackViewerStyleParam(viewerStyle, "margin", c(.01, .05, .01, .05))
for(i in 1:2){
    setTrackYaxisParam(trackList[[i]], "main", FALSE)
}
## Adjust the limit of y-axis
setTrackStyleParam(trackList[[1]], "ylim", c(0, 25))
setTrackStyleParam(trackList[[2]], "ylim", c(-25, 0))
viewTracks(trackList, gr=gr, viewerStyle=viewerStyle)
plot data with y-axis in right side

plot data with y-axis in right side

Adjust the label of y-axis

The style of y-axis can be changed by setting the ylabgp slot in the style of each track.

setTrackStyleParam(trackList[[1]], "ylabgp", list(cex=.8, col="green"))
## set cex to avoid automatic adjust
setTrackStyleParam(trackList[[2]], "ylabgp", list(cex=.8, col="blue"))
setTrackStyleParam(trackList[[2]], "marginBottom", .2)
viewTracks(trackList, gr=gr, viewerStyle=viewerStyle)
plot data with adjusted color and size of y-axis label

plot data with adjusted color and size of y-axis label

The y-axis label can be put at the top or the bottom of each track.

setTrackStyleParam(trackList[[1]], "ylabpos", "bottomleft")
setTrackStyleParam(trackList[[2]], "ylabpos", "topright")
setTrackStyleParam(trackList[[2]], "marginTop", .2)
viewTracks(trackList, gr=gr, viewerStyle=viewerStyle)
plot data with adjusted y-axis label position

plot data with adjusted y-axis label position

For each transcript, the transcript name can be put on the upstream or downstream of the transcript.

trackListN <- trackList
setTrackStyleParam(trackListN[[3]], "ylabpos", "upstream")
setTrackStyleParam(trackListN[[4]], "ylabpos", "downstream")
## set cex to avoid automatic adjust
setTrackStyleParam(trackListN[[3]], "ylabgp", list(cex=.6))
setTrackStyleParam(trackListN[[4]], "ylabgp", list(cex=.6))
gr1 <- range(unlist(GRangesList(sapply(trs, function(.ele) .ele$dat))))
start(gr1) <- start(gr1) - 2000
end(gr1) <- end(gr1) + 2000
viewTracks(trackListN, gr=gr1, viewerStyle=viewerStyle)
plot data with adjusted transcripts name position

plot data with adjusted transcripts name position

Adjust the track color

The track color can be changed by setting the color slot in the style of each track. The first color is for the dat slot of track and the second color is for the dat2 slot.

setTrackStyleParam(trackList[[1]], "color", c("green", "black"))
setTrackStyleParam(trackList[[2]], "color", c("black", "blue"))
for(i in 3:length(trackList)) 
    setTrackStyleParam(trackList[[i]], "color", "black")
viewTracks(trackList, gr=gr, viewerStyle=viewerStyle)
plot data with adjusted track color

plot data with adjusted track color

Adjust the track height

The track height can be changed by setting the height slot in the style of each track. However, the total height for all the tracks should be 1.

trackListH <- trackList
setTrackStyleParam(trackListH[[1]], "height", .1)
setTrackStyleParam(trackListH[[2]], "height", .44)
for(i in 3:length(trackListH)){
    setTrackStyleParam(trackListH[[i]], "height", 
                       (1-(0.1+0.44))/(length(trackListH)-2))
}
viewTracks(trackListH, gr=gr, viewerStyle=viewerStyle)
plot data with adjusted track height

plot data with adjusted track height

Change the track names

The track names such as gene model names can be edited easily by changing the names of trackList.

names(trackList) <- c("cpsf160", "fox2", rep("HSPA8", 5))
viewTracks(trackList, gr=gr, viewerStyle=viewerStyle)
change the track names

change the track names

Show paired data in the same track

trackViewer can be used to show to-be-compared data in the same track side by side.

cpsf160 <- importScore(file.path(extdata, "cpsf160.repA_-.wig"),
                       file.path(extdata, "cpsf160.repB_-.wig"),
                       format="WIG")
strand(cpsf160$dat) <- strand(cpsf160$dat2) <- "-"
setTrackStyleParam(cpsf160, "color", c("black", "red"))
viewTracks(trackList(trs, cpsf160), gr=gr, viewerStyle=viewerStyle)
show two data in the same track

show two data in the same track

Flip the x-axis

The x-axis can be horizotally flipped for the genes in the negative strand.

viewerStyleF <- viewerStyle
setTrackViewerStyleParam(viewerStyleF, "flip", TRUE)
setTrackViewerStyleParam(viewerStyleF, "xaxis", TRUE)
setTrackViewerStyleParam(viewerStyleF, "margin", c(.1, .05, .01, .01))
vp <- viewTracks(trackList, gr=gr, viewerStyle=viewerStyleF)
addGuideLine(c(122929767, 122929969), vp=vp)
addArrowMark(list(x=122929650,
                  y=2),
             label="label",
             col="blue",
             vp=vp)
show data in the flipped track

show data in the flipped track

Optimize the theme

Currently, we support two themes: bw (black and white) and col (colored).

optSty <- optimizeStyle(trackList(repA, fox2, trs), theme="bw")
trackList <- optSty$tracks
viewerStyle <- optSty$style
vp <- viewTracks(trackList, gr=gr, viewerStyle=viewerStyle)
balck & white theme

balck & white theme

optSty <- optimizeStyle(trackList(repA, fox2, trs), theme="col")
trackList <- optSty$tracks
viewerStyle <- optSty$style
vp <- viewTracks(trackList, gr=gr, viewerStyle=viewerStyle)
colorful theme

colorful theme

optSty <- optimizeStyle(trackList(repA, fox2, trs), theme="safe")
trackList <- optSty$tracks
viewerStyle <- optSty$style
vp <- viewTracks(trackList, gr=gr, viewerStyle=viewerStyle)
colorful theme

colorful theme

Plot with breaks

We could plot the tracks with breaks by setting multiple genomic ranges.

gr.breaks <- GRanges("chr11", 
                     IRanges(c(122929275, 122929575, 122929775), 
                             c(122929555, 122929725, 122930122)), 
                     strand="-", percentage=c(.4, .2, .4))
vp <- viewTracks(trackList, gr=gr.breaks, viewerStyle=viewerStyle)
axis with breaks

axis with breaks

Generate and edit an interactive plot using the browseTracks function

As shown above, figures produced by trackViewer are highly customizable, allowing users to alter the label, symbol, color, and size with various functions.

For users who prefer to modify the look and feel of a figure interactively, they can use the function browseTracks to draw interactive tracks, leveraging the htmlwidgets package.

browseTracks(trackList, gr=gr)

interactive tracks

The videos at https://youtu.be/lSmeTu4WMlc and https://youtu.be/lvF0tnJiHQI illustrate how to generate and modify an interactive plot. Please note that the interactive feature is only fully implemented in version 1.19.14 or later.

Operators

If you are interested in drawing a combined track from two input tracks, e.g, adding or substractiong one from the other, then you can try one of the operators such as + and - as showing below.

newtrack <- repA
## Must keep the same format for dat and dat2
newtrack <- parseWIG(newtrack, "chr11", 122929275, 122930122)
newtrack$dat2 <- newtrack$dat
newtrack$dat <- fox2$dat2
setTrackStyleParam(newtrack, "color", c("blue", "red"))
viewTracks(trackList(newtrack, trs), 
           gr=gr, viewerStyle=viewerStyle, operator="+")
show data with operator "+"

show data with operator “+”

viewTracks(trackList(newtrack, trs), gr=gr, viewerStyle=viewerStyle, operator="-")
show data with operator "-"

show data with operator “-”

Alternatively, you can try GRoperator before viewing tracks.

newtrack$dat <- GRoperator(newtrack$dat, newtrack$dat2, col="score", operator="-")
newtrack$dat2 <- GRanges()
viewTracks(trackList(newtrack, trs), gr=gr, viewerStyle=viewerStyle)
show data with operator "-"

show data with operator “-”

Lolliplot

Lolliplot is for the visualization of the methylation/variant/mutation data.

SNP <- c(10, 12, 1400, 1402)
sample.gr <- GRanges("chr1", IRanges(SNP, width=1, names=paste0("snp", SNP)))
features <- GRanges("chr1", IRanges(c(1, 501, 1001), 
                                    width=c(120, 400, 405),
                                    names=paste0("block", 1:3)))
lolliplot(sample.gr, features)

## More SNPs
SNP <- c(10, 100, 105, 108, 400, 410, 420, 600, 700, 805, 840, 1400, 1402)
sample.gr <- GRanges("chr1", IRanges(SNP, width=1, names=paste0("snp", SNP)))
lolliplot(sample.gr, features)

## Define the range
lolliplot(sample.gr, features, ranges = GRanges("chr1", IRanges(104, 109)))

Change the lolliplot color

Change the color of the features.

features$fill <- c("#FF8833", "#51C6E6", "#DFA32D")
lolliplot(sample.gr, features)

Change the color of the lollipop.

sample.gr$color <- sample.int(6, length(SNP), replace=TRUE)
sample.gr$border <- sample(c("gray80", "gray30"), length(SNP), replace=TRUE)
lolliplot(sample.gr, features)

Change the opacity of the lollipop.

sample.gr$alpha <- sample(100:255, length(SNP), replace = TRUE)/255
lolliplot(sample.gr, features)

Add the index labels in the node

sample.gr$label <- as.character(1:length(sample.gr))
sample.gr$label.col <- ifelse(sample.gr$alpha>0.5, "white", "black")
lolliplot(sample.gr, features)

Change the height of the features

features$height <- c(0.02, 0.05, 0.08)
lolliplot(sample.gr, features)

## Specifying the height and its unit
features$height <- list(unit(1/16, "inches"), 
                     unit(3, "mm"), 
                     unit(12, "points"))
lolliplot(sample.gr, features)

Plot multiple transcripts in the features

The metadata ‘featureLayerID’ are used for drawing features in different layers.

features.mul <- rep(features, 2)
features.mul$height[4:6] <- list(unit(1/8, "inches"),
                                 unit(0.5, "lines"),
                                 unit(.2, "char"))
features.mul$fill <- c("#FF8833", "#F9712A", "#DFA32D", 
                       "#51C6E6", "#009DDA", "#4B9CDF")
end(features.mul)[5] <- end(features.mul[5])+50
features.mul$featureLayerID <- 
    paste("tx", rep(1:2, each=length(features)), sep="_")
names(features.mul) <- 
    paste(features.mul$featureLayerID, 
          rep(1:length(features), 2), sep="_")
lolliplot(sample.gr, features.mul)

## One name per transcript
names(features.mul) <- features.mul$featureLayerID
lolliplot(sample.gr, features.mul)

Change the height of a lollipop plot

#Note: the score value is an integer less than 10
sample.gr$score <- sample.int(5, length(sample.gr), replace = TRUE)
lolliplot(sample.gr, features)

##Remove y-axis
lolliplot(sample.gr, features, yaxis=FALSE)

#Try a score value greater than 10
sample.gr$score <- sample.int(20, length(sample.gr), replace=TRUE)
lolliplot(sample.gr, features)

#Try a float numeric score
sample.gr$score <- runif(length(sample.gr))*10
lolliplot(sample.gr, features)

# Score should not be smaller than 1
# remove the alpha for following samples
sample.gr$alpha <- NULL

Customize the x-axis label position

xaxis <- c(1, 200, 400, 701, 1000, 1200, 1402) ## define the position
lolliplot(sample.gr, features, xaxis=xaxis)

names(xaxis) <- xaxis # define the labels
names(xaxis)[4] <- "center" 
lolliplot(sample.gr, features, xaxis=xaxis)

Customize the y-axis label position

yaxis <- c(0, 5) ## define the position
lolliplot(sample.gr, features, yaxis=yaxis)

yaxis <- c(0, 5, 10, 15)
names(yaxis) <- yaxis # define the labels
names(yaxis)[3] <- "y-axis" 
lolliplot(sample.gr, features, yaxis=yaxis)

Jitter the label

sample.gr$dashline.col <- sample.gr$color
lolliplot(sample.gr, features, jitter="label")

Add a legend

legend <- 1:6 ## legend fill color
names(legend) <- paste0("legend", letters[1:6]) ## legend labels
lolliplot(sample.gr, features, legend=legend)

## use list to define more attributes. see ?grid::gpar to get more details.
legend <- list(labels=paste0("legend", LETTERS[1:6]), 
               col=palette()[6:1], 
               fill=palette()[legend])
lolliplot(sample.gr, features, legend=legend)

## if you have multiple tracks, please try to set the legend by list.
## see more examples in the section [Plot multiple samples](#plot-multiple-samples)
legend <- list(legend)
lolliplot(sample.gr, features, legend=legend)

# from version 1.21.8, users can also try to set legend 
# as a column name in the metadata of GRanges.
sample.gr.newlegend <- sample.gr
sample.gr.newlegend$legend <- LETTERS[sample.gr$color]
lolliplot(sample.gr.newlegend, features, legend="legend")

Control the labels

Users can control the paramters of labels by naming the metadata start as label.parameter such as label.parameter.rot or label.parameter.gp. The parameter is used for grid.text.

sample.gr.rot <- sample.gr
sample.gr.rot$label.parameter.rot <- 45
lolliplot(sample.gr.rot, features, legend=legend)

sample.gr.rot$label.parameter.rot <- 60
sample.gr.rot$label.parameter.gp <- gpar(col="brown")
lolliplot(sample.gr.rot, features, legend=legend)

If you want to change the text in the ylab, please try to set the labels in the ylab. Please note that lolliplot does not support any parameters to set the title and xlab. If you want to add the title and xlab, please try to add them by grid.text.

lolliplot(sample.gr.rot, features, legend=legend, ylab="y label here")
grid.text("label of x-axis here", x=.5, y=.01, just="bottom")
grid.text("title here", x=.5, y=.98, just="top", 
          gp=gpar(cex=1.5, fontface="bold"))

Users can control the labels one by one by setting label.parameter.gp. Please note that for each label, the label.parameter.gp must be a list.

label.parameter.gp.brown <- gpar(col="brown")
label.parameter.gp.blue <- gpar(col="blue")
label.parameter.gp.red <- gpar(col="red")
sample.gr$label.parameter.gp <- sample(list(label.parameter.gp.blue,
                                            label.parameter.gp.brown,
                                            label.parameter.gp.red),
                                       length(sample.gr), replace = TRUE)
lolliplot(sample.gr, features)

Change the lolliplot type

Change the shape for “circle” plot

## shape must be "circle", "square", "diamond", "triangle_point_up", or "triangle_point_down"
available.shapes <- c("circle", "square", "diamond", 
                      "triangle_point_up", "triangle_point_down")
sample.gr$shape <- sample(available.shapes, size = length(sample.gr), replace = TRUE)
sample.gr$legend <- paste0("legend", as.numeric(factor(sample.gr$shape)))
lolliplot(sample.gr, features, type="circle", legend = "legend")

Google pin

lolliplot(sample.gr, features, type="pin")

sample.gr$color <- lapply(sample.gr$color, function(.ele) c(.ele, sample.int(6, 1)))
sample.gr$border <- sample.int(6, length(SNP), replace=TRUE)
lolliplot(sample.gr, features, type="pin")

Flag

sample.gr.flag <- sample.gr
sample.gr.flag$label <- names(sample.gr) ## move the names to metadata:label
names(sample.gr.flag) <- NULL
lolliplot(sample.gr.flag, features, 
          ranges=GRanges("chr1", IRanges(0, 1600)), ## use ranges to leave more space on the right margin.
          type="flag")

## change the flag rotation angle
sample.gr.flag$label.rot <- 15
sample.gr.flag$label.rot[c(2, 5)] <- c(60, -15)
sample.gr.flag$label[7] <- "I have a long name"
lolliplot(sample.gr.flag, features, 
          ranges=GRanges("chr1", IRanges(0, 1600)), 
          type="flag")

Pie plot

sample.gr$score <- NULL ## must be removed, because pie will consider all the numeric columns except column "color", "fill", "alpha", "shape", "lwd", "id" and "id.col".
sample.gr$label <- NULL
sample.gr$label.col <- NULL
x <- sample.int(100, length(SNP))
sample.gr$value1 <- x
sample.gr$value2 <- 100 - x
## the length of the color should be no less than that of value1 or value2
sample.gr$color <- rep(list(c("#87CEFA", '#98CE31')), length(SNP))
sample.gr$border <- "gray30"
lolliplot(sample.gr, features, type="pie")

Plot multiple samples

Multiple layers

SNP2 <- sample(4000:8000, 30)
x2 <- sample.int(100, length(SNP2), replace=TRUE)
sample2.gr <- GRanges("chr3", IRanges(SNP2, width=1, names=paste0("snp", SNP2)), 
             value1=x2, value2=100-x2)
sample2.gr$color <- rep(list(c('#DB7575', '#FFD700')), length(SNP2))
sample2.gr$border <- "gray30"

features2 <- GRanges("chr3", IRanges(c(5001, 5801, 7001), 
                                    width=c(500, 500, 405),
                                    names=paste0("block", 4:6)),
                    fill=c("orange", "gray30", "lightblue"),
                    height=unit(c(0.5, 0.3, 0.8), "cm"))
legends <- list(list(labels=c("WT", "MUT"), fill=c("#87CEFA", '#98CE31')), 
                list(labels=c("WT", "MUT"), fill=c('#DB7575', '#FFD700')))
lolliplot(list(A=sample.gr, B=sample2.gr), 
          list(x=features, y=features2), 
          type="pie", legend=legends)

Different layouts are also possible.

sample2.gr$score <- sample2.gr$value1 ## The circle layout needs the score column 
lolliplot(list(A=sample.gr, B=sample2.gr), 
          list(x=features, y=features2), 
          type=c("pie", "circle"), legend=legends)

pie.stack layout

rand.id <- sample.int(length(sample.gr), 3*length(sample.gr), replace=TRUE)
rand.id <- sort(rand.id)
sample.gr.mul.patient <- sample.gr[rand.id]
## pie.stack require metadata "stack.factor", and the metadata can not be 
## stack.factor.order or stack.factor.first
len.max <- max(table(rand.id))
stack.factors <- paste0("patient", formatC(1:len.max, 
                                           width=nchar(as.character(len.max)), 
                                           flag="0"))
sample.gr.mul.patient$stack.factor <- 
    unlist(lapply(table(rand.id), sample, x=stack.factors))
sample.gr.mul.patient$value1 <- 
    sample.int(100, length(sample.gr.mul.patient), replace=TRUE)
sample.gr.mul.patient$value2 <- 100 - sample.gr.mul.patient$value1
patient.color.set <- as.list(as.data.frame(rbind(rainbow(length(stack.factors)), 
                                                 "#FFFFFFFF"), 
                                           stringsAsFactors=FALSE))
names(patient.color.set) <- stack.factors
sample.gr.mul.patient$color <- 
    patient.color.set[sample.gr.mul.patient$stack.factor]
legend <- list(labels=stack.factors, col="gray80", 
               fill=sapply(patient.color.set, `[`, 1))
lolliplot(sample.gr.mul.patient, features, type="pie.stack", 
          legend=legend, dashline.col="gray")

Caterpillar layout

Metadata SNPsideID is used to trigger caterpillar layout. SNPsideID must be ‘top’ or ‘bottom’.

sample.gr$SNPsideID <- sample(c("top", "bottom"), 
                               length(sample.gr),
                               replace=TRUE)
lolliplot(sample.gr, features, type="pie", 
          legend=legends[[1]])

## Two layers
sample2.gr$SNPsideID <- "top"
idx <- sample.int(length(sample2.gr), 15)
sample2.gr$SNPsideID[idx] <- "bottom"
sample2.gr$color[idx] <- '#FFD700'
lolliplot(list(A=sample.gr, B=sample2.gr), 
          list(x=features.mul, y=features2), 
          type=c("pie", "circle"), legend=legends)

EMBL-EBI Proteins API

Following code will show how to use EBI Proteins REST API to get annotations of protein domains.

library(httr) # load library to get data from REST API
APIurl <- "https://www.ebi.ac.uk/proteins/api/" # base URL of the API
taxid <- "9606" # human tax ID
gene <- "TP53" # target gene
orgDB <- "org.Hs.eg.db" # org database to get the uniprot accession id
eid <- mget("TP53", get(sub(".db", "SYMBOL2EG", orgDB)))[[1]]
chr <- mget(eid, get(sub(".db", "CHR", orgDB)))[[1]]
accession <- unlist(lapply(eid, function(.ele){
  mget(.ele, get(sub(".db", "UNIPROT", orgDB)))
}))
stopifnot(length(accession)<=20) # max number of accession is 20

tryCatch({ ## in case the internet connection does not work
  featureURL <- paste0(APIurl, 
                       "features?offset=0&size=-1&reviewed=true",
                       "&types=DNA_BIND%2CMOTIF%2CDOMAIN",
                       "&taxid=", taxid,
                       "&accession=", paste(accession, collapse = "%2C")
  )
  response <- GET(featureURL)
  if(!http_error(response)){
    content <- content(response)
    content <- content[[1]]
    acc <- content$accession
    sequence <- content$sequence
    gr <- GRanges(chr, IRanges(1, nchar(sequence)))
    domains <- do.call(rbind, content$features)
    domains <- GRanges(chr, IRanges(as.numeric(domains[, "begin"]),
                                     as.numeric(domains[, "end"]),
                                     names = domains[, "description"]))
    names(domains)[1] <- "DNA_BIND" ## this is hard coding.
    domains$fill <- 1+seq_along(domains)
    domains$height <- 0.04
    ## GET variations. This part can be replaced by user-defined data.
    variationURL <- paste0(APIurl,
                           "variation?offset=0&size=-1",
                           "&sourcetype=uniprot&dbtype=dbSNP",
                           "&taxid=", taxid,
                           "&accession=", acc)
    response <- GET(variationURL)
    if(!http_error(response)){
      content <- content(response)
      content <- content[[1]]
      keep <- sapply(content$features, function(.ele) length(.ele$evidences)>2 && # filter the data by at least 2 evidences
                       !grepl("Unclassified", .ele$clinicalSignificances)) # filter the data by classified clinical significances.
      nkeep <- c("wildType", "alternativeSequence", "begin", "end",
                 "somaticStatus", "consequenceType", "score")
      content$features <- lapply(content$features[keep], function(.ele){
        .ele$score <- length(.ele$evidences)
        unlist(.ele[nkeep]) 
      })
      variation <- do.call(rbind, content$features)
      variation <- 
        GRanges(chr, 
                IRanges(as.numeric(variation[, "begin"]),
                        width = 1,
                        names = paste0(variation[, "wildType"],
                                       variation[, "begin"],
                                       variation[, "alternativeSequence"])),
                score = as.numeric(variation[, "score"]),
                color = as.numeric(factor(variation[, "consequenceType"]))+1)
      variation$label.parameter.gp <- gpar(cex=.5)
      lolliplot(variation, domains, ranges = gr, ylab = "# evidences", yaxis = FALSE)
    }else{
      message("Can not get variations. error code: ", http_status(response))
    }
  }else{
    message("Can not get features. error code: ", http_status(response))
  }
},error=function(e){
  message(e)
})

Variant Call Format (VCF) data

library(VariantAnnotation)
library(TxDb.Hsapiens.UCSC.hg19.knownGene)
library(org.Hs.eg.db)
fl <- system.file("extdata", "chr22.vcf.gz", package="VariantAnnotation")
gr <- GRanges("22", IRanges(50968014, 50970514, names="TYMP"))
if(.Platform$OS.type!="windows"){# This line is for avoiding error from VariantAnnotation in the windows platform, which will be removed when VariantAnnotation's issue gets fixed.
tab <- TabixFile(fl)
vcf <- readVcf(fl, "hg19", param=gr)
mutation.frequency <- rowRanges(vcf)
mcols(mutation.frequency) <- cbind(mcols(mutation.frequency), 
                                   VariantAnnotation::info(vcf))
mutation.frequency$border <- "gray30"
mutation.frequency$color <- 
    ifelse(grepl("^rs", names(mutation.frequency)), "lightcyan", "lavender")
## Plot Global Allele Frequency based on AC/AN
mutation.frequency$score <- mutation.frequency$AF*100
seqlevelsStyle(mutation.frequency) <- "UCSC" 
}
seqlevelsStyle(gr) <- "UCSC" 
trs <- geneModelFromTxdb(TxDb.Hsapiens.UCSC.hg19.knownGene,
                         org.Hs.eg.db,
                         gr=gr)
features <- c(range(trs[[1]]$dat), range(trs[[5]]$dat))
names(features) <- c(trs[[1]]$name, trs[[5]]$name)
features$fill <- c("lightblue", "mistyrose")
features$height <- c(.02, .04)
if(.Platform$OS.type!="windows"){
lolliplot(mutation.frequency, features, ranges=gr)
}

Methylation data

library(rtracklayer)
session <- browserSession()
query <- ucscTableQuery(session, track="HAIB Methyl RRBS", 
                        range=GRangesForUCSCGenome("hg19", seqnames(gr), ranges(gr)))
tableName(query) <- tableNames(query)[1]
methy <- track(query)
methy <- GRanges(methy)
lolliplot(methy, features, ranges=gr, type="pin")

Change the node size

In the above example, some of the nodes overlap each other. To change the node size, cex prameter could be used.

methy$lwd <- .5
lolliplot(methy, features, ranges=gr, type="pin", cex=.5)

lolliplot(methy, features, ranges=gr, type="circle", cex=.5)

methy$score2 <- max(methy$score) - methy$score
lolliplot(methy, features, ranges=gr, type="pie", cex=.5)

## We can change it one by one
methy$cex <- runif(length(methy))
lolliplot(methy, features, ranges=gr, type="pin")

lolliplot(methy, features, ranges=gr, type="circle")

Change the scale of the x-axis (xscale)

In the above examples, some of the nodes are moved too far from the original coordinates. To rescale, the x-axis could be reset as below.

methy$cex <- 1
lolliplot(methy, features, ranges=gr, rescale = TRUE)

## by set percentage for features and non-features segments
xaxis <- c(50968014, 50968514, 50968710, 50968838, 50970514)
rescale <- c(.3, .4, .3)
lolliplot(methy, features, ranges=gr, type="pin",
          rescale = rescale, xaxis = xaxis)

## by set data.frame to rescale
rescale <- data.frame(
  from.start = c(50968014, 50968515, 50968838), 
  from.end   = c(50968514, 50968837, 50970514),
  to.start   = c(50968014, 50968838, 50969501), 
  to.end     = c(50968837, 50969500, 50970514)
)
lolliplot(methy, features, ranges=gr, type="pin",
          rescale = rescale, xaxis = xaxis)

Dandelion Plot

Sometimes, there are as many as hundreds of SNPs invoved in one gene. Dandelion plot can be used to depict such dense SNPs. Please note that the height of the dandelion indicates the desity of the SNPs.

dandelion.plot(methy, features, ranges=gr, type="pin")

Change the type of Dandelion plot

There are one more type for dandelion plot, i.e., type “fan”. The area of the fan indicates the percentage of methylation or rate of mutation.

methy$color <- 3
methy$border <- "gray"
## Score info is required and the score must be a number in [0, 1]
m <- max(methy$score)
methy$score <- methy$score/m
dandelion.plot(methy, features, ranges=gr, type="fan")

methy$color <- rep(list(c(3, 5)), length(methy))
methy$score2 <- methy$score2/m
legends <- list(list(labels=c("s1", "s2"), fill=c(3, 5)))
dandelion.plot(methy, features, ranges=gr, type="pie", legend=legends)

Change the number of dandelions

## Less dandelions
dandelion.plot(methy, features, ranges=gr, type="circle", maxgaps=1/10)

## More dandelions
dandelion.plot(methy, features, ranges=gr, type="circle", maxgaps=1/100)

Users can also specity the maximum distance between neighboring dandelions by settimg the maxgaps as a GRanges object.

maxgaps <- tile(gr, n = 10)[[1]]
dandelion.plot(methy, features, ranges=gr, type="circle", maxgaps=maxgaps)

Add y-axis (yaxis)

Set yaxis to TRUE to add y-axis, and set heightMethod=mean to use the mean score as the height.

dandelion.plot(methy, features, ranges=gr, type="pie", 
               maxgaps=1/100, yaxis = TRUE, heightMethod = mean,
               ylab='mean of methy scores')

yaxis = c(0, 0.5, 1)
dandelion.plot(methy, features, ranges=gr, type="pie", 
               maxgaps=1/100, yaxis = yaxis, heightMethod = mean,
               ylab='mean of methy scores')

Plot the lollipop plot with the coverage and annotation tracks

gene <- geneTrack(get("HSPA8", org.Hs.egSYMBOL2EG), TxDb.Hsapiens.UCSC.hg19.knownGene)[[1]]
SNPs <- GRanges("chr11", IRanges(sample(122929275:122930122, size = 20), width = 1), strand="-")
SNPs$score <- sample.int(5, length(SNPs), replace = TRUE)
SNPs$color <- sample.int(6, length(SNPs), replace=TRUE)
SNPs$border <- "gray80"
SNPs$feature.height = .1
SNPs$cex <- .5
gene$dat2 <- SNPs
optSty <- optimizeStyle(trackList(repA, fox2, gene), theme="col")
trackList <- optSty$tracks
viewerStyle <- optSty$style
gr <- GRanges("chr11", IRanges(122929275, 122930122))
setTrackStyleParam(trackList[[3]], "ylabgp", list(cex=.8))
vp <- viewTracks(trackList, gr=gr, viewerStyle=viewerStyle)

## lollipopData track
SNPs2 <- GRanges("chr11", IRanges(sample(122929275:122930122, size = 30), width = 1), strand="-")
SNPs2 <- c(SNPs2, promoters(gene$dat, upstream = 0, downstream = 1))
SNPs2$score <- sample.int(3, length(SNPs2), replace = TRUE)
SNPs2$color <- sample.int(6, length(SNPs2), replace=TRUE)
SNPs2$border <- "gray30"
SNPs2$feature.height = .1
SNPs2$cex <- .5
SNPs$cex <- .5
lollipopData <- new("track", dat=SNPs, dat2=SNPs2, type="lollipopData")
gene <- geneTrack(get("HSPA8", org.Hs.egSYMBOL2EG), TxDb.Hsapiens.UCSC.hg19.knownGene)[[1]]
optSty <- optimizeStyle(trackList(repA, lollipopData, gene, heightDist = c(3, 3, 1)), theme="col")
trackList <- optSty$tracks
viewerStyle <- optSty$style
gr <- GRanges("chr11", IRanges(122929275, 122930122))
setTrackStyleParam(trackList[[2]], "ylabgp", list(cex=.8))
vp <- viewTracks(trackList, gr=gr, viewerStyle=viewerStyle)
addGuideLine(122929538, vp=vp)

Ideogram Plot

Plot ideograms with a list of chromosomes and a genome.

ideo <- loadIdeogram("hg38")
dataList <- ideo
dataList$score <- as.numeric(dataList$gieStain)
dataList <- dataList[dataList$gieStain!="gneg"]
dataList <- GRangesList(dataList)
ideogramPlot(ideo, dataList, 
            layout=list("chr1", c("chr3", "chr22"), 
                        c("chr4", "chr21")))

Plot genomic interactions data

Different from most of the available tools, plotGInteractions try to plot the data with the 2D structure. The nodes indicate the region with interactions and the edges indicates the interactions. The size of the nodes are relative to the width of the region. The features could be the enhancers, promoters or genes. The enhancer and promoter are shown as points with symbol 11 and 13.

library(TxDb.Hsapiens.UCSC.hg19.knownGene)
library(InteractionSet)
gi <- readRDS(system.file("extdata", "gi.rds", package="trackViewer"))
range <- GRanges("chr2", IRanges(234500000, 235000000))
feature.gr <- genes(TxDb.Hsapiens.UCSC.hg19.knownGene)
feature.gr <- subsetByOverlaps(feature.gr, regions(gi))
feature.gr$col <- sample(1:7, length(feature.gr), replace=TRUE)
feature.gr$type <- sample(c("promoter", "enhancer", "gene"), 
                          length(feature.gr), replace=TRUE, 
                          prob=c(0.1, 0.2, 0.7))
plotGInteractions(gi, range, feature.gr)

Web application of trackViewer

We created a web application of trackViewer (available in 1.19.14 or later) by leveraging the R package Shiny. The web application of trackViewer and sample data are available at https://github.com/jianhong/trackViewer.documentation/tree/master/trackViewerShinyApp. Here is a demo on how to to use the web application at https://www.nature.com/articles/s41592-019-0430-y#Sec2 Supplementary Video 5.

Session Info

sessionInfo()

R version 3.6.2 (2019-12-12) Platform: x86_64-pc-linux-gnu (64-bit) Running under: Ubuntu 18.04.3 LTS

Matrix products: default BLAS: /home/biocbuild/bbs-3.10-bioc/R/lib/libRblas.so LAPACK: /home/biocbuild/bbs-3.10-bioc/R/lib/libRlapack.so

locale: [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_US.UTF-8 LC_COLLATE=C
[5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=en_US.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

attached base packages: [1] grid parallel stats4 stats graphics grDevices utils
[8] datasets methods base

other attached packages: [1] InteractionSet_1.14.0
[2] httr_1.4.1
[3] VariantAnnotation_1.32.0
[4] Rsamtools_2.2.2
[5] Biostrings_2.54.0
[6] XVector_0.26.0
[7] SummarizedExperiment_1.16.1
[8] DelayedArray_0.12.2
[9] BiocParallel_1.20.1
[10] matrixStats_0.55.0
[11] org.Hs.eg.db_3.10.0
[12] TxDb.Hsapiens.UCSC.hg19.knownGene_3.2.2 [13] GenomicFeatures_1.38.2
[14] AnnotationDbi_1.48.0
[15] Biobase_2.46.0
[16] Gviz_1.30.3
[17] rtracklayer_1.46.0
[18] trackViewer_1.22.1
[19] GenomicRanges_1.38.0
[20] GenomeInfoDb_1.22.0
[21] IRanges_2.20.2
[22] S4Vectors_0.24.3
[23] BiocGenerics_0.32.0

loaded via a namespace (and not attached): [1] ProtGenerics_1.18.0 bitops_1.0-6 bit64_0.9-7
[4] RColorBrewer_1.1-2 progress_1.2.2 Rgraphviz_2.30.0
[7] tools_3.6.2 backports_1.1.5 R6_2.4.1
[10] rpart_4.1-15 Hmisc_4.3-1 DBI_1.1.0
[13] lazyeval_0.2.2 colorspace_1.4-1 nnet_7.3-12
[16] tidyselect_1.0.0 gridExtra_2.3 prettyunits_1.1.1
[19] bit_1.1-15.2 curl_4.3 compiler_3.6.2
[22] graph_1.64.0 htmlTable_1.13.3 grImport_0.9-3
[25] scales_1.1.0 checkmate_2.0.0 askpass_1.1
[28] rappdirs_0.3.1 stringr_1.4.0 digest_0.6.24
[31] foreign_0.8-75 rmarkdown_2.1 dichromat_2.0-0
[34] base64enc_0.1-3 jpeg_0.1-8.1 pkgconfig_2.0.3
[37] htmltools_0.4.0 plotrix_3.7-7 highr_0.8
[40] ensembldb_2.10.2 BSgenome_1.54.0 dbplyr_1.4.2
[43] htmlwidgets_1.5.1 rlang_0.4.4 rstudioapi_0.11
[46] RSQLite_2.2.0 jsonlite_1.6.1 acepack_1.4.1
[49] dplyr_0.8.4 RCurl_1.98-1.1 magrittr_1.5
[52] GenomeInfoDbData_1.2.2 Formula_1.2-3 Matrix_1.2-18
[55] Rcpp_1.0.3 munsell_0.5.0 lifecycle_0.1.0
[58] stringi_1.4.6 yaml_2.2.1 zlibbioc_1.32.0
[61] BiocFileCache_1.10.2 blob_1.2.1 crayon_1.3.4
[64] lattice_0.20-40 splines_3.6.2 hms_0.5.3
[67] knitr_1.28 pillar_1.4.3 biomaRt_2.42.0
[70] XML_3.99-0.3 glue_1.3.1 evaluate_0.14
[73] biovizBase_1.34.1 latticeExtra_0.6-29 data.table_1.12.8
[76] BiocManager_1.30.10 png_0.1-7 vctrs_0.2.3
[79] gtable_0.3.0 openssl_1.4.1 purrr_0.3.3
[82] assertthat_0.2.1 ggplot2_3.2.1 xfun_0.12
[85] AnnotationFilter_1.10.0 survival_3.1-8 tibble_2.1.3
[88] GenomicAlignments_1.22.1 memoise_1.1.0 cluster_2.1.0
[91] BiocStyle_2.14.4