CopyNumberPlots: create copy-number specific plots using karyoploteR

library(knitr)
opts_chunk$set(concordance=FALSE)
knitr::opts_chunk$set(fig.width = 18)
knitr::opts_chunk$set(fig.height = 12)
set.seed(21666641)

Introduction

Data visualisation is a powerful tool used for data analysis and exploration in many fields. Genomics data analysis is one of these fields where good visualisation tools can be of great help. The aim of r BiocStyle::Biocpkg("CopyNumberPlots") is to offer the user an easy way to create copy-number related plots using the infrastructure provided by the R package r BiocStyle::Biocpkg("karyoploteR").

In addition to a set of specialized plotting functions for copy-number analysis data and results (plotBAF, plotCopyNumberCalls, ...), r BiocStyle::Biocpkg("CopyNumberPlots") contains a number of data loading functions to help parsing and loading the results of widely used copy-number calling software such as r BiocStyle::Biocpkg("DNAcopy"), DECoN or CNVkit.

Finally, since r BiocStyle::Biocpkg("CopyNumberPlots") extends the functionality of r BiocStyle::Biocpkg("karyoploteR"), it is possible to combine the plotting functions of both packages to get the perfect figure for your data.

Installation

r BiocStyle::Biocpkg("CopyNumberPlots") is a Bioconductor package and to install it we have to use r BiocStyle::Biocpkg("BiocManager").

  if (!requireNamespace("BiocManager", quietly = TRUE))
      install.packages("BiocManager")
  BiocManager::install("CopyNumberPlots")

We can also install the package from github to get the latest devel version, but beware that it might be incompatible with the release version of Bioconductor!

```` {r, eval = FALSE} BiocManager::install("bernatgel/CopyNumberPlots")

# Quick Start

To start working with `r BiocStyle::Biocpkg("CopyNumberPlots")` we will need
to use the `plotKaryoptype` function from `r BiocStyle::Biocpkg("karyoploteR")`.
If you want more information on how to customize it, use for other organisms
or genome version, etc... you can take a look at the
[karyoploteR tutorial](https://bernatgel.github.io/karyoploter_tutorial/) and 
specifically at the section on 
[how to plot ideograms](https://bernatgel.github.io/karyoploter_tutorial//Tutorial/CreateIdeogram/CreateIdeogram.html).

For this quick start example we'll plot SNP-array data simulating a cancer 
genome. The data is in a file included with the package. You can use almost
any table-like file format, including the Final Report file you would get from
Illumina's Genome Studio. In this case, to keep the example small, we have 
data only for chomosome 1.

To load the data we'll use `loadSNPData` which will detect the right columns, 
read the data and build a GRanges object for us. 

If data uses Ensembl-style chromosome names (1,2,3,...,X,Y) instead of 
default karyoploteR UCSC chromosome names (chr1,chr2,chr3,...,chrX,chrY)
we could change the chromosome style to UCSC with the function `UCSCStyle`.

```r
  library(CopyNumberPlots)

  s1.file <- system.file("extdata", "S1.rawdata.txt", package = "CopyNumberPlots", mustWork = TRUE)
  s1 <- loadSNPData(s1.file)
  s1

Once we have our data loaded we can start plotting. We'll start by creating a karyoplot using plotKaryotype. If we were plotting more than one chromosome, we could use plot.type=4 to get all chromosomes in a single line one next to the other. You can get more information on the available plot types at the karyoploteR tutorial.

  kp <- plotKaryotype(chromosomes="chr1")

And once we have a karyoplot we can start adding out data. We can plot the B-allele frequency using plotBAF

  kp <- plotKaryotype(chromosomes="chr1")
  plotBAF(kp, s1)

We can plot LRR using plotLRR

  kp <- plotKaryotype(chromosomes="chr1")
  plotLRR(kp, s1)

And we can see in this plot that points with a LRR below -4 (and above 2) are plotted in red at -4 (and at 2) so we don't lose them.

We can also use the data positioning parameters r0 and r1 to add more than one data type on the same plot.

  kp <- plotKaryotype(chromosomes="chr1")
  plotBAF(kp, s1, r0=0.55, r1=1)
  plotLRR(kp, s1, r0=0, r1=0.45)

Finally, we can load a copy number calling made on this data and plot it. To load the copy number calls in this file we can use the function loadCopyNumberCalls that will read the data, identify the correct columns and create a GRanges object for us.

  s1.calls.file <- system.file("extdata", "S1.segments.txt", package = "CopyNumberPlots", mustWork = TRUE)
  s1.calls <- loadCopyNumberCalls(s1.calls.file)
  s1.calls

And then use plotCopyNumberCalls to add them to the previous plot.

  kp <- plotKaryotype(chromosomes="chr1")
  plotBAF(kp, s1, r0=0.6, r1=1)
  plotLRR(kp, s1, r0=0.15, r1=0.55)
  plotCopyNumberCalls(kp, s1.calls, r0=0, r1=0.10)

With that the main functionality of r BiocStyle::Biocpkg("CopyNumberPlots") is covered. It is important to take into account that since we are extending the functionality of r BiocStyle::Biocpkg("karyoploteR"), we can use all karyoploteR functions to add more data and other data types into these plots.

In the following pages you will find more information on how to load data to use with r BiocStyle::Biocpkg("CopyNumberPlots"), how to create other plot types and how to customize them.

Loading Copy-Number Data

The plotting functions in r BiocStyle::Biocpkg("CopyNumberPlots") expect data to be in a GRanges with a few columns with specific names:

You can create these structures yourself, but r BiocStyle::Biocpkg("CopyNumberPlots") has functions to help in loading both raw data (mainly SNP-array and aCGH data) and copy-number calls.

Load Raw Data

The main function to load raw data is loadSNPData. It will take either a file or an R object (data.frame or similar) and will load it, detect the columns with the needed information (chromosome, position, log-ratio, B-allele frequency) based on the column names and build a GRanges object ready to use by the plotting functions.

  raw.data.file <- system.file("extdata", "snp.data_test.csv", package = "CopyNumberPlots", mustWork=TRUE)
  snps <- loadSNPData(raw.data.file)
  snps

When run, the function will tell us the columns it identified and will proceed load the data. To identify the columns it will internally use a set of regular expressions that work in most cases including on the 'Final Report' files created by Illumina's Genome Studio. If for any reason the automatic identification of the columns failed, it is possible to specify the exact column names using the appropiate parameters (chr.col, start.col, end.col...).

Load Copy-Number Calls

Another set of functions included in the package are functions to load the results of copy-number calling algorithms, the copy number calls per se. In this case we also have a generic function, loadCopyNumberCalls, and a few functions specialized in specific copy-number calling packages.

Again, the generic function can work with a file or an R object with a table-like structure and will try to discover the right columns itself. It will return a GRanges with the copy-number called segments and the optional columns cn for integer copy-number values, loh for loss-of-heterozigosity regions and segment.value for values computed for the segments (for example, mean value of the probes in the segment).

As an example we will generate a "random" calling

  cn.data <- toGRanges(c("chr14:66459785-86459774", "chr17:68663111-88866308",
                         "chr10:43426998-83426994", "chr3:88892741-120892733",
                         "chr2:12464318-52464316", "chrX:7665575-27665562"))

  cn.data$CopyNumberInteger <- sample(c(0,1,3,4), size = 6, replace = TRUE)
  cn.data$LossHetero <- cn.data$CopyNumberInteger<2

  cn.data

and load it

  cn.calls <- loadCopyNumberCalls(cn.data)
  cn.calls

we can see how the columns for cn and loh were correctly identified.

To plot this objet we can call, for example plotCopyNumberCalls.

  kp <- plotKaryotype(plot.type = 1)
  plotCopyNumberCalls(kp, cn.calls = cn.calls)

There are other specialized functions that will load either the R object produced by copy-number calling R packages or the files produced by either R or external copy-number calling software.

Currently there are specilized functions to load the data produced by:

Plotting Copy-Number Data

Once we have data loaded (or directly created by us) we can plot it.

There are two functions to plot raw data (plotBAF and plotLRR) and three functions to plot the copy-number calls (plotCopyNumberCalls, plotCopyNumberCallsAsLines and plotCopyNumberSummary).

Plotting Raw Data

To demonstrate the raw-data plotting functions we'll use two example files included with the package

  s1.file <- system.file("extdata", "S1.rawdata.txt", package = "CopyNumberPlots", mustWork = TRUE)
  s1 <- loadSNPData(s1.file)
  head(s1)

  s2.file <- system.file("extdata", "S2.rawdata.txt", package = "CopyNumberPlots", mustWork = TRUE)
  s2 <- loadSNPData(s2.file)
  head(s2)

plotBAF

To plot the B-Allele frequency (BAF) we'll use plotBAF. We'll start creating a karyoplot using karyoploteR's plotKaryotype and then add the BAF values into it.

  kp <- plotKaryotype(chromosomes="chr1")
  plotBAF(kp, snps=s1)

We can change a number of parameters to alter the appearance of the plot. We can activate and deactivate the axis and label, we can change the color, size and glyph (shape) of the points, we can use r0 and r1 alter the vertical position of the data and in general we can use any of the standard base R plotting parameters.

  kp <- plotKaryotype(chromosomes="chr1")
  plotBAF(kp, snps=s1, r0=0, r1=0.2, labels = "BAF1", points.col = "orange",
          points.cex = 2, points.pch = 4, axis.cex = 0.3)
  plotBAF(kp, snps=s1, r0=0.3, r1=0.5, labels = "BAF2", points.col = "red",
          points.cex = 0.5, points.pch = 8, axis.cex = 0.7)
  plotBAF(kp, snps=s1, r0=0.6, r1=1, labels = "BAF3", 
          points.col = "#FF552222", points.cex = 1.8, points.pch = 16, 
          axis.cex = 0.7)

If we want to plot more than one sample, if we have the data in a list of GRanges or in a GRanges list, plotBAF will take care of it and plot the different samples one below the other. It will also use the names of the list as labels to identify the different samples.

  samples <- list("Sample1"=s1, "Sample2"=s2)
  kp <- plotKaryotype(chromosomes="chr1")
  plotBAF(kp, snps=samples)

Plot LRR

The function plotLRR is equivalent to the plotBAF function but will plot the data in the "lrr" column.

  kp <- plotKaryotype(chromosomes="chr1")
  kpAddBaseNumbers(kp)
  plotLRR(kp, snps=s1)

plotLRR has a few specific parameters. Since the range of the data points is not limited to [0,1] as in BAF, you can define the ymin and ymax values and any point falling out of the [ymin, ymax] range will be plotted in red within this range.

This can help us identify out-of-range data, such as the deletion arround 50Mb in the plot above or the gained region at ~220Mb.

Changing the values of ymin and ymax we can see a bit different picture

  kp <- plotKaryotype(chromosomes="chr1")
  kpAddBaseNumbers(kp)
  plotLRR(kp, snps=s1,  ymin=-1.5, ymax=1.5)

In this case we see many more points out-of-range. We can change the appearance of this points, changing their color, for example, of we can change how they are represented, using a density plot instead of raw points.

  kp <- plotKaryotype(chromosomes="chr1")
  kpAddBaseNumbers(kp)
  plotLRR(kp, snps=s1,  ymin=-1.5, ymax=1.5, out.of.range = "density")

In this case, due to the very few points in the example, the default parameters for the density plot are not optimal. We can increase the window size to compute the density using larger windows. For example, we can set the window to 1 megabase.

  kp <- plotKaryotype(chromosomes="chr1")
  plotLRR(kp, snps=s1,  ymin=-1.5, ymax=1.5, out.of.range = "density", density.window = 1e6)

And we can see the peaks corresponding to the accumulation of out-of-range points.

Finally, we can control the presence and color of the horizontal line marking the 0 with the "line.at.0.*" parameters.

We can also use the standard customization options with plotLRR.

  kp <- plotKaryotype(chromosomes="chr1")
  plotLRR(kp, snps=s1, r0=0, r1=0.2, labels = "LRR1", points.col = "orange",
          points.cex = 2, points.pch = 4, axis.cex = 0.3)
  plotLRR(kp, snps=s1, r0=0.3, r1=0.5, labels = "LRR2", points.col = "red",
          points.cex = 0.5, points.pch = 8, axis.cex = 0.7, ymin=-1.5, ymax=1.5,
          out.of.range.col = "gold", out.of.range = "density",
          density.window = 10e6, density.height = 0.3)
  plotLRR(kp, snps=s1, r0=0.6, r1=1, labels = "LRR3",
          points.col = "#FF552222", points.cex = 1.8, points.pch = 16,
          axis.cex = 0.7)

Plot Copy-Number Calls

The final data type we can plot with r BiocStyle::Biocpkg("CopyNumberPlots") are copy number calls, that is, the results from copy-number calling algorithms. To plot that we need a GRanges object with a at least one column of: * "cn" for integer copy number calls * "segment.value" for non-integer segment regional values * "loh" a logical for loss-of-heterozygosity

As an example we'll use the data generated by ASCAT in a cancer cell line.

  s1.calls.file <- system.file("extdata", "S1.segments.txt", package = "CopyNumberPlots", mustWork = TRUE)
  s1.calls <- loadCopyNumberCalls(s1.calls.file)
  s2.calls <- loadCopyNumberCalls(system.file("extdata", "S2.segments.txt", package = "CopyNumberPlots", mustWork = TRUE))
  s3.calls <- loadCopyNumberCalls(system.file("extdata", "S3.segments.txt", package = "CopyNumberPlots", mustWork = TRUE))
  s1.calls

plotCopyNumberCalls

The first function to plot the copy-number calls is plotCopyNumberCalls, which will plot them as colored rectangles over the genome. It will create 2 lines of rectangles: the top one with copy-number values and the bottom one with loss-of-heterozygosity in blue.

  kp <- plotKaryotype(chromosomes = "chr1")
  plotCopyNumberCalls(kp, s1.calls)

By default we'll see losses in green, 2n regions in gray and gains in yellow-orange-red. And the LOH regions as a blue line below the CN data. We can change the colors used with cn.colors. This parameter will take any value accepted by getCopyNumberColors, including the predefined palletes. You can find them all in the documentation of getCopyNumberColors. This fuction can also help us creating a legend.

  kp <- plotKaryotype(chromosomes="chr1")
  plotCopyNumberCalls(kp, s1.calls, cn.colors = "red_blue", loh.color = "orange", r1=0.8)
  cn.cols <- getCopyNumberColors(colors = "red_blue")
  legend("top", legend=names(cn.cols), fill = cn.cols, ncol=length(cn.cols))

As with the other plotting functions, giving it a list of GRanges will plot them all.

  cn.calls <- list("Sample1"=s1.calls, "Sample2"=s2.calls, "Sample3"=s3.calls)
  kp <- plotKaryotype(chromosomes="chr1")
  plotCopyNumberCalls(kp, cn.calls, r1=0.3)

plotCopyNumberCallsAsLines

Another option is to plot the copy-number calls as lines using the function plotCopyNumberCallsAsLines. We'll show a single chromosome in this case.

  kp <- plotKaryotype(chr="chr1")
  plotCopyNumberCallsAsLines(kp, s1.calls)

In this case we can change the standard customization options and make it use segments instead of lines using the additional parameter style.

  kp <- plotKaryotype(chr="chr1")
  plotCopyNumberCallsAsLines(kp, s1.calls, style = "segments")

plotCopyNumberSummary

Finally, to plot a view of the accumulation of copy number alterations we can use plotCopyNumberSummary. It will create a coverage plot of gains and losses over all samples in our dataset.

  cn.cols <- getCopyNumberColors(colors = "green_orange_red")
  kp <- plotKaryotype(chromosomes="chr1")
  kpDataBackground(kp, color = cn.cols["2"], r0=0.3)
  plotCopyNumberCalls(kp, cn.calls, loh.height = 0, r0=0.3)
  plotCopyNumberSummary(kp, cn.calls, r1=0.25)

And we can change the appearance of the summary using the direction parameter.

  cn.cols <- getCopyNumberColors(colors = "green_orange_red")
  kp <- plotKaryotype(chromosomes="chr1")
  kpDataBackground(kp, color = cn.cols["2"], r0=0.3)
  plotCopyNumberCalls(kp, cn.calls, loh.height = 0, r0=0.3)
  plotCopyNumberSummary(kp, cn.calls, r1=0.25, direction = "out")

Session Info

  sessionInfo()


Try the CopyNumberPlots package in your browser

Any scripts or data that you put into this service are public.

CopyNumberPlots documentation built on Nov. 8, 2020, 6:51 p.m.