Visualize Data
There are numerous ways to visualize data within toxEval that can serve different objectives. Each function described below provides a unique method for exploring EAR data that can be customized to fit a user’s specific study objectives. All of these functions rely on EAR values computed by the get_chemical_summary function.
Most of the graphing and table functions have an argument “mean_logic”. This allows the user to decided if the mean (TRUE) of the site or the maximum (FALSE) should be displayed. It depends on the type of analysis that is being done on which value will be more appropriate. “mean_logic” defaults to FALSE. If the sites have the same number of samples, and the goal is to explore a representative view of the data, than using TRUE would be more appropriate.
plot_tox_boxplots
The plot_tox_boxplots function creates a set of boxplots representing EAR values computed from the original input data that was modified by the processing steps above, and dependent on the choice of several input options. See “Summarizing the data” for a description of how the EAR values are computed, aggregated, and summarized. Choosing “Chemical Class” in the category argument will generate separate boxplots for each unique class. “Chemical” will generate boxplots for each individual chemical, and “Biological” will generate boxplots for each group in the selected ToxCast annotation.
# Other options:
# plot_tox_boxplots(chemical_summary, "Chemical Class")
# plot_tox_boxplots(chemical_summary, "Chemical") It is also possible to display a threshold line using the hit_threshold argument. The graph will then include the number of sites with detections, the threshold line, and the number of “hits” indicating how many sites that have EAR values exceeding the hit_threshold.
The graph shows a slightly different result for a single site. In the example below, a subset of data from the Maumee River is defined that will be used throughout this document to facilitate single site examples.
library(dplyr)
maumee <- filter(chemical_summary, shortName == "Maumee")
maumee_site <- filter(tox_list$chem_site, `Short Name` == "Maumee")In this example, the function plot_tox_boxplots is used with the subset of data from the Maumee River. For this single site graph, the number of chemicals that were detected and have associated endpoint ACCs represented are displayed.
See “Box Plot Details” for a description of what the lines and dots represent in these box plots.
plot_tox_stacks
The plot_tox_stacks function creates a set of stacked bar charts representing EAR values computed from the original input data that was modified by the processing steps above, and dependent on the choice of several input options. See “Summarizing the data” for a description on how the EAR values are computed, aggregated, and summarized. Choosing “Chemical Class” in the category argument will generate separate stacked bars for each unique class. “Chemical” will generate stacked bars for each individual chemical, and “Biological” will generate stacked bars for each group in the selected ToxCast annotation. There is an option include_legend to turn on and off the legend. It may be impractical for instance to show the legend for “Chemical” if there are hundreds of chemicals.
# More options:
# plot_tox_stacks(chemical_summary,
# chem_site = tox_list$chem_site,
# category = "Chemical Class")
# plot_tox_stacks(chemical_summary,
# chem_site = tox_list$chem_site,
# category = "Chemical", include_legend = FALSE)The graph shows a slightly different result for a single site. The graph now shows each individual sample. The EAR values are no longer the mean or maximum at the site, but rather the sums of the individual EAR values.
plot_tox_heatmap
The plot_tox_heatmap function creates a heat (tile) map with sites on the x-axis, a specified grouping on the y-axis (defined by the category argument), and color shading defining the mean or maximum EAR. See “Summarizing the data” for a description on how the EAR values are computed, aggregated, and summarized. The y-axis grouping can be “Biological”,“Chemical Class”, or “Chemical”. When specifying the “Chemical” option, a secondary y-axis is automatically included to group chemicals into chemical class. The function computes default breaks for the color scale to match the spread of the data, but breaks can also be customized with the breaks argument.
This is a function where it may be ideal to create a custom order to the sites (for example, west-to-east). See the above section “Custom configuration” for instructions on how to convert the character vector sites to a factor with ordered levels.
plot_tox_endpoints
The plot_tox_endpoints function creates a set of boxplots representing EAR values for each endPoint based on a selected filter. The data is first filtered down to what group is specified in the filterBy argument. The filterBy argument must match one of the unique options in the category. For example, if the category is “Chemical Class”, then the filterBy argument must be one of the defined “Chemical Class” options such as “Herbicide”.
After the data is filtered, a boxplot is generated for each endPoint. The EAR values that are used to create the boxplots are the mean or maximum (as defined by mean_logic) for each site as described in “Summarizing the data”.
plot_tox_endpoints(chemical_summary, top_num = 10,
category = "Biological",
filterBy = "Cell Cycle")
# More options:
# plot_tox_endpoints(chemical_summary,
# category = "Chemical Class",
# filterBy = "PAHs")
# plot_tox_endpoints(chemical_summary,
# category = "Chemical",
# filterBy = "Atrazine")See “Box Plot Details” for a description of what the lines and dots represent in these box plots.
Box Plot Details
The following diagram describes the lines and dots within the boxplots:
How to save a plot?
The ggsave function can be used to save plots. This technique does have an advantage by taking away much of the guessing game for how big or resolved the plot should be. For RStudio users, the size by default is determined by how large the “Plots” tab is sized. This could be handy when sharing plots with colleagues during project analysis. An example of using the ggsave function is:
