This document describes the use of the functions included in the conStruct package for visualizing analysis outputs. For more information on how to run a
conStruct analysis, see the companion vignette for running conStruct.
Throughout, this vignette will make use of the example data output objects generated by a
make.figs is set to
TRUE in a
conStruct run, the run will finish by calling the function
make.all.the.plots. As the name implies, this function makes all the relevant plots from a set of conStruct results:
More information is available in the documentation for the function, which you can view using the command:
If you deleted the output plots from an analysis, or if you set
FALSE to avoid making them in the first place, you can make them by calling the
make.all.the.plots function. The arguments you have to specify are a
conStruct.results output object and a
data.block output object, both of which are automatically generated and saved when you execute a
conStruct analysis. You must also specify a
prefix, which will be prepended to all output pdf file names. If you choose, you can specify a the colors you want each layer to be plotted in; if none are specified, the function will use its own internal vector of colors, which I think look nice but are otherwise arbitrary.
An example call to
make.all.the.plots using the example output data objects loaded above is shown below.
Generally, users are most interested in the estimated admixture proportions for each sample. These are commonly visualized using STRUCTURE plots and pie plots. Functions for both are included in the package, and their use is detailed below.
Probably the most common method for visualizing admixture proportions is using a stacked bar plot (commonly called a STRUCTURE plot after the model-based clustering method
Users can generate a STRUCTURE plot for their data using the command
make.structure.plot, (see documentation at
help(make.structure.plot)). This function takes as its principal argument the estimated admixture proportions and makes a STRUCTURE plot in the plotting window. An example is given below.
First, we load the
conStruct.results data output object and, for convenience, assign the maximum a posteriori admixture parameter estimates to a variable with a shorter name:
load("my_conStruct.results.Robj") # assign the MAP admixture proportions from # the first MCMC chain to a variable # with a new name admix.props <- conStruct.results$chain_1$MAP$admix.proportions
Now we can visualize the results:
# make a STRUCTURE plot using the # maximum a posteriori (MAP) estimates # from the first chain of a conStruct run make.structure.plot(admix.proportions = admix.props)
The function also includes a variety of options for tweaking the order of the plotted samples.
# re-order the stacking order of the layers make.structure.plot(admix.proportions = admix.props, layer.order = c(2,1,3), sort.by = 2)
# provide a custom sample ordering # in this case by sample latitude make.structure.plot(admix.proportions = admix.props, sample.order = order(data.block$coords[,2]))
# add sample names make.structure.plot(admix.proportions = admix.props, sample.names = row.names(data.block$coords), mar = c(4.5,4,2,2))
It is often also useful to visualize estimated admixture proportions in a spatial context by plotting them on a map. The most common way to do this is to plot a pie plot at the sampling location of each sample, in which each modeled layer gets its own slice of the pie (
K wedges), and the size of each slice in the pie is proportional to the sample’s admixture proportion in that layer.
Users can make an admixture pie plot with their own data using the command
make.admix.pie.plot (see documentation at
help(make.admix.pie.plot). This function takes as its principal arguments the estimated admixture proportions and the sample coordinates, then makes an admixture pie plot in the plotting window. An example is given below:
# make an admix pie plot using the # maximum a posteriori (MAP) estimates # from the first chain of a conStruct run make.admix.pie.plot(admix.proportions = admix.props, coords = data.block$coords)
# increase pie chart size make.admix.pie.plot(admix.proportions = admix.props, coords = data.block$coords, radii = 4)
# zoom in on a subsection of the map make.admix.pie.plot(admix.proportions = admix.props, coords = data.block$coords, x.lim = c(-130,-120), y.lim = c(49,56))
Users can also add the pie plot directly to a map of their own creation using the
make.admix.pie.plot by setting the
add argument to
# add pie plot to an existing map # make the desired map maps::map(xlim = range(data.block$coords[,1]) + c(-5,5), ylim = range(data.block$coords[,2])+c(-2,2), col="gray") # add the admixture pie plot make.admix.pie.plot(admix.proportions = admix.props, coords = data.block$coords, add = TRUE)
If you’ve run multiple
conStruct analyses you may want to visually compare them. Although you could always just open up both sets of output pdfs, label-switching between independent runs can make visual comparisons difficult. Label-switching different models have the same, or very similar, estimated admixture proportions, but with a different permutation of layer labels (e.g., Layer 1 in run 1, and Layer 3 in run 2). To enable easy comparison between a pair of
conStruct runs, you can use the function
To do so, you need to specify to sets of
conStruct.results output R objects, as well as the
data.block objects associated with each run. Independent runs with the same model can be compared, as can analyses run with different models (e.g., spatial vs. nonspatial) or different values of
K. The only restriction is that if the user is comparing two models run with different values of
K, the run with the smaller value should be specified first (
conStruct.results2). Documentation for
compare.two.runs can be found using the command
help(compare.two.runs). Example usage is shown below:
# load output files from a run with # the spatial model and K=4 load("spK4.conStruct.results.Robj") load("spK4.data.block.Robj") # assign to new variable names spK4_cr <- conStruct.results spK4_db <- data.block # load output files from a run with # the spatial model and K=3 load("spK3.conStruct.results.Robj") load("spK3.data.block.Robj") # assign to new variable names spK3_cr <- conStruct.results spK3_db <- data.block # compare the two runs compare.two.runs(conStruct.results1=spK3_cr, data.block1=spK3_db, conStruct.results2=spK4_cr, data.block2=spK4_db, prefix="spK3_vs_spK4") # generates a bunch of pdf figures