plot_structure_map: Plot STRUCTURE like results on a map.

In hemstrow/snpR: Whole-Genome Analysis Tools for Use with Single Nucleotide Polymorphism Data

Plot STRUCTURE like results on a map.

Description

Plots the mean cluster assignment for each population on a map using the scatterpie package alongside any additional simple feature objects (see sf from the sf package). Assignments must be given in the format provided by plot_structure. This function is a wrapper which sources code from github.

Usage

plot_structure_map(
  assignments,
  k,
  facet,
  pop_coordinates,
  layers = NULL,
  pop_names = T,
  viridis.option = "viridis",
  alt.palette = NULL,
  radius_scale = 0.05,
  label_args = NULL,
  crop = FALSE,
  scale_bar = list(),
  compass = list(style = ggspatial::north_arrow_fancy_orienteering, location = "br"),
  ask = TRUE
)

Arguments

assignments	Structure like results, parsed in or generated via plot_structure, which generates the needed plot data.
k	numeric. Value of K (number of clusters) to plot.
facet	character. The facet by which data is broken down in the passed assignments.
pop_coordinates	sf object, see the documentation for sf function from the sf package. sf object containing points/coordinates for each facet level. Must contain a column of data with population labels named identically to the provided facet (for example, named "pop" if "pop" is the provided facet.)
layers	list of ggplot2 layer objects, default NULL. Additional ggplot layers to be plotted in order, below the pie charts, such as maps with borders, temperatures, forest cover, etc. As a special note, link[ggnewscale]{new_scale_fill} can be used to add additional fill aesthetic layers without conflict from the resulting pie charts. See examples.
pop_names	logical, default T. If true, facet level names will be displayed on the map.
viridis.option	character, default "viridis". Viridis color scale option. See scale_gradient for details.
alt.palette	character or NULL, default NULL. Optional palette of colors to use instead of viridis palette the pie charts.
radius_scale	numeric 0-1, default 0.05. Scale for pie chart radii as a proportion of the total map space.
label_args	list, default NULL. Named list of arguments passed to geom_label_repel. For example, passing list(max.overlaps = 14) will add the max.overlaps argument to the function call.
crop	logical, default F. If TRUE, will will crop the plot around the sample points. If false will show the full extent of the data, often set by any additional sf objects being plotted.
scale_bar	list or NULL, default list(). Arguments passed to the annotation_scale function from ggspatial to add a scale to the plot. If NULL, no scale added.
compass	list or NULL, default list(style = ggspatial::north_arrow_fancy_orienteering, location = "br"). Arguments passed to annotation_north_arrow function from ggspatial to add a compass to the plot. If NULL, no compass added. Note that calls to alternative styles, like north_arrow_fancy_orienteering cannot have () after them, like they would if called directly.
ask	logical, default TRUE. Should the function ask for confirmation before sourcing github code?

Details

Currently, this only works for simple, sample specific facets. Coordinates for pie charts should be provided as an sf object, where one column, named for the facet being plotted, provides the subfacet level names matching those in the assignments. Additional sf objects can be provided, which will also be plotted. Note that there is no need to standardize the CRS across the objects, since each will be transformed to match the sample coordinates.

Examples

## Not run: 
# get an sf of the sampling locations
lat_long <- data.frame(SMR = c(44.365931, -121.140420), 
                       CLF = c(44.267718, -121.255805), 
                       OPL = c(44.485958, -121.298360), 
                       ASP = c(43.891693, -121.448360), 
                       UPD = c(43.891755, -121.451600), 
                       PAL = c(43.714114, -121.272797)) # coords for points
lat_long <- t(lat_long)
colnames(lat_long) <- c("lat", "long")
lat_long <- as.data.frame(lat_long)
lat_long$pop <- rownames(lat_long)
psf <- sf::st_as_sf(as.data.frame(lat_long), coords = c("long", "lat"))
psf <- sf::`st_crs<-`(psf, "EPSG:4326")

# get the assignments (STRUCTURE-like results)
assignments <- plot_structure(stickSNPs, "pop", alpha = 1) 

# get a map of oregon as a background from the maps package. 
background <- rnaturalearth::ne_states(iso_a2 = "US", returnclass = "sp")
background <- sf::st_as_sf(background)
background <- background[background$name %in% "Oregon",]

# make the plot
plot_structure_map(assignments, k = 2, facet = "pop", pop_coordinates = psf,
                   layers = list(ggplot2::geom_sf(data = background, 
                                     ggplot2::aes(fill = "example")),
                   ggnewscale::new_scale_fill()), radius_scale = .2)

## End(Not run)

hemstrow/snpR documentation built on March 20, 2024, 7:03 a.m.