Web-based interactive charts (using D3.js) for the analysis of
experimental crosses to identify genetic loci (quantitative trait
loci, QTL) contributing to variation in quantitative traits.
Broman (2015)
R/qtlcharts is an R package to create interactive charts for QTL data, for use with R/qtl. [website]
It is built with D3, using a set of reusable panels (also available separately, as d3panels).
For example charts, see the R/qtlcharts website.
Install R/qtlcharts from CRAN using
install.packages("qtlcharts")
Alternatively, install it from its GitHub repository. You first need to install the R/qtl, htmlwidgets, and devtools packages.
install.packages(c("qtl", "htmlwidgets", "devtools"))
Then install R/qtlcharts using the install_github
function in the
devtools package.
library(devtools)install_github("kbroman/qtlcharts")
Try the following example, which creates an interactive chart with LOD curves linked to estimated QTL effects.
library(qtl)library(qtlcharts)data(hyper)hyper <- calc.genoprob(hyper, step=1)out <- scanone(hyper)iplotScanone(out, hyper)
Also try iplotCorr
, an image of a correlation matrix (for the
gene expression of a set of 100 genes) linked to the underlying
scatterplots, with the points in the scatterplot colored by their
genotype at a QTL:
library(qtlcharts)data(geneExpr)iplotCorr(geneExpr$expr, geneExpr$genotype)
Finally, try iboxplot
, a plot of the quantiles of many
distributions, linked to the underlying histograms.
library(qtlcharts)# simulate some datan.ind <- 500n.gene <- 10000expr <- matrix(rnorm(n.ind * n.gene, (1:n.ind)/n.ind*3), ncol=n.gene)dimnames(expr) <- list(paste0("ind", 1:n.ind), paste0("gene", 1:n.gene))# generate the plotiboxplot(expr)
The R/qtlcharts package as a whole is distributed under [GPL-3 (GNU General Public License version 3)[https://www.gnu.org/licenses/gpl-3.0.en.html].
R/qtlcharts incorporates the following other open source software components, which have their own license agreements.
Fix bug in idotplot()
so that it works with a single group.
Fix bug in iplotMap()
so that the drop-down menu works in Firefox.
Update to use CoffeeScript v 2.2.2. Using babel to compile to ES5.
Switch from bower to yarn for javascript dependency management.
Revise scat2scat()
so that the chart options xlab2
and ylab2
can be vectors of length of scat2data
, so that each dataset can
have different x- and y-axis labels. Default is to take them from
the column names of the datasets in scat2data
.
All charts can now take a heading
and a footer
within
chartOpts
. The former will make an <h2>
heading above the
figure; the latter will make a <div>
at the bottom. These are
intended for stand-alone html files.
In the caption
chart option, now using .html()
rather than
.text()
so that you can insert a bit of html (such as <b>
or
<code>
).
caption
within chartOpts
which will
show up as text below the figure. This is intended for stand-alone
html files.Added new tool for exploring pleiotropy between two traits,
ipleiotropy()
.
Fixed X chromosome case in iplotScantwo, so that only relevant two-locus genotypes are shown.
Added new chart option pointsize
for iplotCorr
.
In the coffeescript functions, ensure that list arguments like
margin
have all of the necessary components. This avoid the
problem of everything being messed up if for example margin
is
specified within defining margin.inner
.
lodcharts
now includes a quantitative scale for position on the
x-axis in the case of a single chromosome
In iplotScanone
and iplotScantwo
, in phenotype x genotype plots
when switching between markers on the same chromosome, animate the
movement of the points rather than destroying and re-creating the
panel.
Rewrite underlying javascript to use new version of d3panels.
Add a new chart, scat2scat
. The idea is to summarize each of a
long series of scatterplots with a pair of numbers. Then a
scatterplot of those summary statistics is linked to the underlying
details: click on a point in the main scatterplot and have the
underlying scatterplot be shown.
Add a new chart, itriplot
, for plotting trinomial probabilities,
represented as points in an equilateral triangle.
Refactor iplotPXG
and idotplot
so that idotplot
is the main
function, and iplotPXG
calls it.
Add some additional options, such as horizontal
for iplotMap
and iplotPXG
.
Change the name of some options, such as linecolor
and linewidth
in iplotCurves
(in place of strokecolor
and strokewidth
).
Add a digits
argument for all plot functions, with the aim to
reduce the size of the datasets included in the resulting charts.
Removed the vignettes from the package (for complicated reasons); they're available at the R/qtlcharts website.
Fix proliferation of tool tips
For use with Shiny, clear SVG before drawing
Changed license and author list in order to post the package on CRAN, http://cran.r-project.org
idotplot
and iplot
now use the names of the input data as
individual IDs if indID
is missing or NULL
.
iplotScanone
with pxgtype="ci"
. In the case of
phenotypes with missing values, the confidence intervals were
incorrect.iplotCorr
has argument scatterplots
that controls whether
scatterplots will be shown when clicking on pixels in the heatmap.
(If scatterplots=FALSE
, we skip the scatterplots.)iplotMScanone
can plot just points (rather than curves) for the
LOD scores and QTL effects in the lower and right-hand panels.
Fix a bug in iplotMScanone
(x-axis labels in right-hand plot
weren't being shown)
setScreenSize
function for controlling the default sizes of
charts.Added idotplot
function for plotting a quantitative variable in
different categories. (It's just like iplotPXG
, but with data
values not connected to a cross object.)
Reorganized the d3panels code: just
using d3panels.min.js
and d3panels.min.css
rather than linking
to js code for individual panels.
Refactored the entire package to use htmlwidgets. A big advantage is that the charts now work nicely within RStudio.
To save a plot to a file, you now need to assign the result of a plot
function to an object and then use the
htmlwidgets function saveWidget
.
library(qtlcharts)library(qtl)data(hyper)hyper <- calc.genoprob(hyper, step=1)out <- scanone(hyper)chart <- iplotScanone(out, hyper)htmlwidgets::saveWidget(chart, file="hyper_scanone.html")
It's now simpler to include interactive charts within an R Markdown
document. You simply make the plots in a code chunk in the way that
you would at the R prompt. There's no longer a need to worry about
print_qtlcharts_resources()
or results="asis"
.
Separated out the basic panel functions as a separate repository, d3panels, to make it easier for them to be used separately from R/qtlcharts.