bmdplotwithgradient: BMD plot with color gradient

View source: R/bmdplotwithgradient.R

bmdplotwithgradientR Documentation

BMD plot with color gradient


Provides an ECDF plot of BMD values with a horizontal color gradient coding, for each item, for the theoretical signal as a function of the dose (concentration). The idea is to display the amplitude and the intensity of the response of each item on the BMD ECDF plot, in addition to the BMD ordered values. This plot is of interest especially when not too much items are presented. To maximize the lisibility of the plot, one can manually pre-select items based on its own criteria (e.g. functional group of interest).


bmdplotwithgradient(extendedres, BMDtype = c("zSD", "xfold"),
                                   xmin, xmax, y0shift = TRUE, scaling = TRUE,
                                   facetby, facetby2, 
                                   shapeby, npoints = 50, 
                                   line.size, point.size = 1,
                                   ncol4faceting, limits4colgradient,
                                   lowercol = "darkblue", uppercol = "darkred",
                                   add.label, label.size = 2,
                                   BMD_log_transfo = TRUE)



the dataframe of results provided by bmdcalc (res) or a subset of this data frame (selected lines). This dataframe can be extended with additional columns coming for example from the functional annotation of items, and some lines can be replicated if their corresponding item has more than one annotation. This extended dataframe must at least contain the column giving the BMD values (BMD.zSD or BMD.xfold depending of chosen BMDtype), identification of each curve (id), the column model naming the fitted model and the values of the parameters (columns b, c, d, e, f).


The type of BMD to plot, "zSD" (default choice) or "xfold".


Optional minimal dose/concentration for definition of the x range.


Optional maximal dose/concentration for definition of the x range (can be defined as max(f$omicdata$dose) with f the output of drcfit() for example).


If TRUE (default choice) for each item the signal is shifted to have the theoretical signal at the control at 0.


If TRUE, default choice, for each item the signal is shifted to have the theoretical signal at the control at 0 and scaled by dividing by the maximal absolute signal change (up or down) from the signal at the control maxychange.


optional argument naming the column of extendedres chosen to split the plot in facets using ggplot2::facet_wrap (no split if omitted).


optional argument naming the column of extendedres chosen as an additional argument to split the plot in facets using ggplot2::facet_grid, with columns defined by facetby and rows defined by facetby2 (no split if omitted).


optional argument naming the column of extendedres chosen to shape the BMD points (no difference if shapeby if omitted).


Number of points computed on each curve in order to define the signal color gradient (= number of doses or concentrations for which the theoretical signal is computed from the fitted model for each item).


Size of the horizontal lines for plotting each signal color gradient.


Size of the BMD points.


Number of columns for facetting (not used if facetby2 is also provided.


Optional vector giving minimal and maximal value of the signal for the color gradient.


Chosen color for the lower values of the signal.


Chosen color for the upper values of the signal.


Points are replaced by labels of items if TRUE.


Size of labels if add.label is TRUE.


If TRUE, the default option, a log transformation of the BMD is used in the plot. This option cannot be used with a null value of xmin in input.


BMD values are plotted as an ECDF plot, as with plot.bmdcalc using "ecdf" as plottype. In addition is plotted an horizontal color gradient for each item coding for the signal level at each dose (or concentration). The optional use of columns to code for shape and/or facets for each item is particularly intended to give a view of all the dose-response per group (e.g. metabolic pathways). Those groups must be coded in a column of extendedres. In case where one item is allocated to more than one group during the annotation process, the line of this item must be replicated in extendedres as many times as the number of annotation groups in which it was allocated.

For each item of the extended dataframe, the name of the model (column model) and the values of the parameters (columns b, c, d, e, f) are used to compute theoretical dose-response curves, and so the corresponding signal color gradient, in the range [xmin ; xmax].


a ggplot object.


Marie-Laure Delignette-Muller

See Also

See plot.bmdcalc and plot.bmdboot.


# (1) 
# A toy example on a very small subsample of a microarray data set
datafilename <- system.file("extdata", "transcripto_very_small_sample.txt", package="DRomics")

o <- microarraydata(datafilename, check = TRUE, norm.method = "cyclicloess")
s_quad <- itemselect(o, select.method = "quadratic", FDR = 0.01)
f <- drcfit(s_quad, progressbar = TRUE)
r <- bmdcalc(f)

# Plot of all the BMD values with color dose-response gradient 
bmdplotwithgradient(r$res, BMDtype = "zSD")  

# Same plot without signal scaling
bmdplotwithgradient(r$res, BMDtype = "zSD", scaling = FALSE) 

# Plot of all the BMD values with color dose-response gradient 
# with definition of xmax from the maximal tested dose
bmdplotwithgradient(r$res, BMDtype = "zSD",
                    xmax = max(f$omicdata$dose)) 

# Add of item labels
bmdplotwithgradient(r$res, BMDtype = "zSD",
                    xmax = max(f$omicdata$dose), add.label = TRUE) 

# The same plot in raw scale (we can fix xmin at 0 in this case) 
bmdplotwithgradient(r$res, BMDtype = "zSD", xmin = 0,
                    BMD_log_transfo = FALSE) 

# The same plot in log scale with defining xmin and xmax at a chosen values 
bmdplotwithgradient(r$res, BMDtype = "zSD",
                    xmin =  min(f$omicdata$dose[f$omicdata$dose != 0] / 2), 
                    xmax = max(f$omicdata$dose),
                    BMD_log_transfo = TRUE) 

# Plot of all the BMD values with color dose-response gradient 
# faceted by response trend and shaped by model
bmdplotwithgradient(r$res, BMDtype = "zSD",
                      facetby = "trend", shapeby = "model") 
# same plot changing the names of the facets 
levels(r$res$trend) <- c("bell shape", "decreasing", "increasing", "U shape")

bmdplotwithgradient(r$res, BMDtype = "zSD",
                      facetby = "trend", shapeby = "model")

# same plot changing the labels of the legends 
# and inversing the two guides
if (require(ggplot2)) bmdplotwithgradient(r$res, BMDtype = "zSD",
                      facetby = "trend", shapeby = "model") + 
                      labs(col = "signal value", shape = "model") +
                      guides(colour = guide_colourbar(order = 1),
                      shape = guide_legend(order = 2))
# (2)
# Plot of BMD values with color dose-response gradient
# faceted by metabolic pathway (from annotation of the selected items)
# and shaped by dose-response trend

# An example from the paper published by Larras et al. 2020
# in Journal of Hazardous Materials
# A example of plot obtained with this function is in Figure 5 in Larras et al. 2020

# the dataframe with metabolomic results (output $res of bmdcalc() or bmdboot() functions)
resfilename <- system.file("extdata", "triclosanSVmetabres.txt", package="DRomics")
res <- read.table(resfilename, header = TRUE, stringsAsFactors = TRUE)

# the dataframe with annotation of each item identified in the previous file
# each item may have more than one annotation (-> more than one line)
annotfilename <- system.file("extdata", "triclosanSVmetabannot.txt", package="DRomics")
annot <- read.table(annotfilename, header = TRUE, stringsAsFactors = TRUE)

# Merging of both previous dataframes
# in order to obtain an extenderes dataframe
extendedres <- merge(x = res, y = annot, by.x = "id", by.y = "metab.code")

### (2.a) BMDplot with gradient by pathway
bmdplotwithgradient(extendedres, BMDtype = "zSD",
                     facetby = "path_class", 
                     xmax = 7.76, # maximal tested dose in those data
                       shapeby = "trend") 

### (2.a) BMDplot with gradient by pathway without scaling
bmdplotwithgradient(extendedres, BMDtype = "zSD",
                     facetby = "path_class", xmax = 7.76,
                       shapeby = "trend", scaling = FALSE) 

# (2.b) BMDplot with gradient by pathway
# forcing the limits of the colour gradient at other 
# values than observed minimal and maximal values of the signal
bmdplotwithgradient(extendedres, BMDtype = "zSD",
                      facetby = "path_class", 
                       shapeby = "trend",
                       limits4colgradient = c(-1, 1)) 

# (2.c) The same example changing the gradient colors and the line size
bmdplotwithgradient(extendedres, BMDtype = "zSD",
                      facetby = "path_class", 
                       shapeby = "trend",
                       line.size = 3, 
                       lowercol = "darkgreen", uppercol = "orange") 

# (2.d) The same example with only lipid metabolism pathclass
# and identification of the metabolites
LMres <- extendedres[extendedres$path_class == "Lipid metabolism", ]
bmdplotwithgradient(LMres, BMDtype = "zSD",
                       line.size = 3, 
                       add.label = TRUE, label.size = 3) 

# (3) 
# An example on a microarray data set (a subsample of a greater data set)
datafilename <- system.file("extdata", "transcripto_sample.txt", package="DRomics")

(o <- microarraydata(datafilename, check = TRUE, norm.method = "cyclicloess"))
(s_quad <- itemselect(o, select.method = "quadratic", FDR = 0.001))
(f <- drcfit(s_quad, progressbar = TRUE))
(r <- bmdcalc(f))

bmdplotwithgradient(r$res, BMDtype = "zSD",
                      facetby = "trend", 
                      shapeby = "model") 
# without scaling
bmdplotwithgradient(r$res, BMDtype = "zSD",
                      scaling = FALSE, facetby = "trend", 
                      shapeby = "model") 

aursiber/DRomics documentation built on June 19, 2024, 5:48 p.m.