Time Profile Plots • ospsuite.plots

1. Introduction

This vignette documents and illustrates workflows for producing Time Profile plots using the ospsuite.plots library.

Time profile plots are commonly used to compare observed and simulated data over time. In these plots, observed data are typically represented as scatter points with error bars indicating population range or confidence intervals, while simulated data are displayed using lines with shaded ribbons representing population range or confidence intervals.

Basic documentation of the function can be found using: ?plotTimeProfile. The output of the function is a ggplot object.

1.1 Setup

This vignette uses the ospsuite.plots and tidyr libraries. We will use the default settings of ospsuite.plots (see vignette(“ospsuite.plots”, package = “ospsuite.plots”)) but will adjust the legend position.

library(ospsuite.plots)
library(tidyr)

# Set Defaults
oldDefaults <- ospsuite.plots::setDefaults()

# Adjust legend position for better aesthetics
theme_update(legend.position = "top")
theme_update(legend.direction = "vertical")
theme_update(legend.box = "horizontal")
theme_update(legend.title = element_blank())

1.2 Example Data

This vignette uses randomly generated example datasets provided by the package:

1.2.1 Simulated and Observed Data

The following datasets are used in the example:

simData1 and obsData1: ‘exponential decay’
simData2 and obsData2: ‘first order absorption with exponential decay’
simData and obsData: combination of simData1 and simData2 / obsData1 and obsData2

simData1 <- exampleDataTimeProfile %>%
  dplyr::filter(SetID == "DataSet1") %>%
  dplyr::filter(Type == "simulated") %>%
  dplyr::select(c("time", "values", "minValues", "maxValues", "caption"))

simData2 <- exampleDataTimeProfile %>%
  dplyr::filter(SetID == "DataSet2") %>%
  dplyr::filter(Type == "simulated") %>%
  dplyr::select(c("time", "values", "minValues", "maxValues", "caption"))

simData <- rbind(simData1, simData2)

obsData1 <- exampleDataTimeProfile %>%
  dplyr::filter(SetID == "DataSet1") %>%
  dplyr::filter(Type == "observed") %>%
  dplyr::select(c("time", "values", "sd", "maxValues", "minValues", "caption"))

obsData2 <- exampleDataTimeProfile %>%
  dplyr::filter(SetID == "DataSet2") %>%
  dplyr::filter(Type == "observed") %>%
  dplyr::select(c("time", "values", "sd", "maxValues", "minValues", "caption"))

obsData <- rbind(obsData1, obsData2)

knitr::kable(head(simData), digits = 3, caption = "First rows of example data simData")

First rows of example data simData
time	values	minValues	maxValues	caption
0.0	15.000	12.000	18.000	Simulated Data 1
0.1	14.851	11.857	17.857	Simulated Data 1
0.2	14.703	11.715	17.714	Simulated Data 1
0.3	14.557	11.576	17.573	Simulated Data 1
0.4	14.412	11.438	17.433	Simulated Data 1
0.5	14.268	11.301	17.294	Simulated Data 1

simDataLloq and obsDataLloq: dataset with a column defining LLOQ.

simDataLloq <- exampleDataTimeProfile %>%
  dplyr::filter(SetID == c("DataSet3")) %>%
  dplyr::filter(Type == "simulated") %>%
  dplyr::filter(dimension == "concentration") %>%
  dplyr::select(c("time", "values", "caption"))

obsDataLloq <- exampleDataTimeProfile %>%
  dplyr::filter(SetID == "DataSet3") %>%
  dplyr::filter(Type == "observed") %>%
  dplyr::filter(dimension == "concentration") %>%
  dplyr::select(c("time", "values", "caption", "lloq", "error_relative"))

simData2Dimension: dataset where the “values” column has mixed dimensions: “concentration” and “fraction”.

simData2Dimension <- exampleDataTimeProfile %>%
  dplyr::filter(SetID == "DataSet3") %>%
  dplyr::filter(Type == "simulated") %>%
  dplyr::select(c("time", "values", "dimension", "caption"))

obsData2Dimension <- exampleDataTimeProfile %>%
  dplyr::filter(SetID == "DataSet3") %>%
  dplyr::filter(Type == "observed") %>%
  dplyr::select(c("time", "values", "dimension", "caption", "lloq", "error_relative"))

obsDataGender: observed dataset with gender information, and simDataGender: a mean model presentation.

simDataGender <- exampleDataTimeProfile %>%
  dplyr::filter(SetID == "DataSet4") %>%
  dplyr::filter(Type == "simulated") %>%
  dplyr::select(c("time", "values", "caption"))

obsDataGender <- exampleDataTimeProfile %>%
  dplyr::filter(SetID == "DataSet4") %>%
  dplyr::filter(Type == "observed") %>%
  dplyr::select(c("time", "values", "caption", "gender"))

1.2.2 MetaData

Metadata is a list that contains dimension and unit information for dataset columns. If available, axis labels are set by this information. If a time unit can be identified for the x-axis, breaks are set according to this unit (see ?updateScaleArgumentsForTimeUnit).

metaData <- attr(exampleDataTimeProfile, "metaData")
knitr::kable(metaData2DataFrame(metaData), digits = 2, caption = "List of meta data")

List of meta data
	time	values
dimension	Time	Concentration
unit	h	mg/l

2 Examples

The following sections demonstrate how to plot a Time Profile for specific scenarios.

2.1 Plot Simulated Data Only

2.1.1 Basic Example with Multiple Simulations

Datasets mapped to data are displayed as lines. The aesthetic groupby, mapped in the example to the column caption, groups profiles by the caption column. This means caption is internally mapped to all aesthetics defined in the variable groupAesthetics. By default, these are color, linetype, shape (only relevant for observed data), and fill.

plotTimeProfile(
  data = simData,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    groupby = caption
  )
)

Time profile plot showing multiple simulated concentration curves over time. Different simulation profiles are displayed as colored lines, with each line representing a different simulation scenario grouped by caption. The plot demonstrates basic time profile visualization for simulated data only.

2.1.2 Multiple Simulations with Confidence Interval

Mapping ymin and ymax will add a ribbon to the time profile, indicating a prediction confidence interval or population variance.

plotTimeProfile(
  data = simData,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    ymin = minValues,
    ymax = maxValues,
    groupby = caption
  )
)

Time profile plot showing simulated concentration curves with confidence intervals. Multiple simulation profiles are displayed as colored lines with shaded ribbon areas representing confidence intervals (ymin to ymax). This demonstrates how to visualize uncertainty or population variance in simulated time profile data.

2.2 Plot Observed Data Only

2.2.1 Basic Example with Multiple Observed Data Sets

A dataset mapped to observed data is displayed as points.

plotTimeProfile(
  observedData = obsData,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    groupby = caption
  )
)

Time profile plot showing observed data points over time. Multiple observed datasets are displayed as colored scatter points, with each dataset grouped by caption. The plot demonstrates basic time profile visualization for observed data without error bars or confidence intervals.

2.2.2 Observed Data Sets with Confidence Interval

Mapping ymin and ymax adds error bars.

plotTimeProfile(
  observedData = obsData,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    ymin = minValues,
    ymax = maxValues,
    groupby = caption
  )
)

Time profile plot showing observed data points with confidence intervals. Multiple observed datasets are displayed as colored scatter points with vertical error bars representing confidence intervals (ymin to ymax). This demonstrates how to visualize measurement uncertainty in observed time profile data.

2.2.3 Usage of Aesthetic “Error”

The dataset includes a column with the standard deviation sd. If mapped to ‘error’, this variable will be used to create corresponding ymin and ymax values for the error bars ymin = values - sd and ymax = values + sd. If yscale = 'log', ymin values below 0 are set to y. Additionally, error_relative can be used where a multiplicative error is assumed: ymin = values / error_relative and ymax = values * error_relative.

plotTimeProfile(
  observedData = obsData,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    error = sd,
    groupby = caption
  )
)

Time profile plot showing observed data points with error bars calculated from standard deviation. The plot displays colored scatter points with vertical error bars where ymin = values - sd and ymax = values + sd, demonstrating the use of the 'error' aesthetic for automatic error bar calculation.

2.2.4 Observed Data with LLOQ

If lloq is mapped to a column indicating the lower limit of quantification, a horizontal line for the lloq values is displayed, and all values below lloq are plotted with decreased alpha. As the comparison is done by row, multiple lloq values are possible.

plotTimeProfile(
  observedData = obsDataLloq,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    groupby = caption,
    error_relative = error_relative,
    groupby = caption
  )
)
#> Warning: Duplicated aesthetics after name standardisation:
#> groupby

Time profile plot showing observed data with lower limit of quantification (LLOQ) handling. The plot displays scatter points with a horizontal LLOQ line, where data points below the LLOQ threshold are shown with reduced transparency and relative error bars are included.

2.2.5 Omit Data Points Flagged as Missing Dependent Variable (MDV)

The following code adds a new column where all values higher than 10 are flagged as mdv. This leads to a plot without any observed data points higher than 10 (removing the first observation).

mdvData <- obsData
mdvData$mdv <- mdvData$values > 10

plotTimeProfile(
  observedData = mdvData,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    ymin = minValues,
    ymax = maxValues,
    groupby = caption
  )
)

Time profile plot demonstrating missing dependent variable (MDV) exclusion. Observed data points are plotted with confidence intervals, but data points with values greater than 10 are excluded from the visualization, showing how the MDV flag can be used to filter out specific data points.

2.3 Plot Simulated and Observed Data

By plotting simulated and observed data together, you can often find pairs of corresponding datasets. This can be done using a common legend entry (see section 2.3.1) or by defining a mapping table, where each observed dataset is mapped to one simulated dataset (see section 2.3.2). There may also be examples with independent datasets (see section 2.3.3).

2.3.1 Corresponding Simulated and Observed Datasets with Common Legend Entry

In this example, we create a new column with a common caption for simulated and observed data. This column is then mapped to the aesthetic groupby, leading to a single common legend for both observed and simulated data.

# Create datasets with common caption
simData <- data.frame(simData) %>%
  dplyr::mutate(captionCommon = gsub("Simulated ", "", caption))

obsData <- data.frame(obsData) %>%
  dplyr::mutate(captionCommon = gsub("Observed ", "", caption))

plotTimeProfile(
  data = simData,
  observedData = obsData,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    ymin = minValues,
    ymax = maxValues,
    groupby = captionCommon
  )
)

Time profile plot combining simulated and observed data with shared legend entries. Simulated data appear as colored lines with ribbon confidence intervals, while observed data appear as matching colored points with error bars. Both datasets share common legend entries by having matching caption names.

2.3.2 Corresponding Simulated and Observed Datasets with Separate Legend Entries and Mapping Table

In this example, we create a mapping table with one column for observed and one column for simulated. This table is passed to the function as the input variable mapSimulatedAndObserved.

mapSimulatedAndObserved <- data.frame(
  simulated = unique(simData$caption),
  observed = unique(obsData$caption)
)

knitr::kable(mapSimulatedAndObserved)

simulated	observed
Simulated Data 1	Observed Data 1
Simulated Data 2	Observed Data 2


plotTimeProfile(
  data = simData,
  observedData = obsData,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    ymin = minValues,
    ymax = maxValues,
    groupby = caption
  ),
  mapSimulatedAndObserved = mapSimulatedAndObserved
)

Time profile plot combining simulated and observed data with separate legend entries using mapping table. Simulated data appear as colored lines with ribbons, observed data as points with error bars. The mapping table pairs specific simulated datasets with corresponding observed datasets, creating separate legend entries for each.

If not all simulated datasets have corresponding observed datasets (or vice versa), it is possible to fill the mapping table with empty strings for the missing datasets. The empty strings should be at the end of the table.

mapSimulatedAndObserved <- data.frame(
  simulated = unique(simData$caption),
  observed = c(unique(obsData1$caption), "")
)

knitr::kable(mapSimulatedAndObserved)

simulated	observed
Simulated Data 1	Observed Data 1
Simulated Data 2


plotTimeProfile(
  data = simData,
  observedData = obsData1,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    ymin = minValues,
    ymax = maxValues,
    groupby = caption
  ),
  mapSimulatedAndObserved = mapSimulatedAndObserved
)

Time profile plot demonstrating partial mapping between simulated and observed datasets. Shows how to handle cases where not all simulated datasets have corresponding observed datasets, using empty strings in the mapping table for missing pairings.

2.3.3 Independent Simulated and Observed Datasets

The example below shows individual observed data compared to one simulation. Here, a mapping between observed and simulated data doesn’t make sense. The groupby aesthetic is used to group the observed data by color, fill, and shape. To avoid an extra color for the simulated line, the line color is set by geomLineAttributes as an attribute and not as an aesthetic. The simulated line inherits the linetype aesthetic from the groupby aesthetic, adding a linetype legend for the simulated data.

plotTimeProfile(
  data = simDataGender,
  observedData = obsDataGender,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    groupby = caption
  ),
  geomLineAttributes = list(color = "black")
) +
  theme(legend.position = "right") +
  labs(
    color = "Observed Data",
    shape = "Observed Data",
    fill = "Observed Data",
    linetype = "Simulation"
  ) +
  theme(legend.title = element_text())
#> Ignoring unknown labels:
#> • linetype : "Simulation"

Time profile plot combining multiple simulated and observed datasets with customized legend formatting. Shows simulated data as black lines and observed data as colored points, with separate legends for simulation and observed data components, and custom legend positioning and titles.

2.3.4 Multiple Simulations and Observed Data Sets without Legends

To map groupby with an empty variable groupAesthetics leads to a plot without legends.

plotTimeProfile(
  data = simData,
  observedData = obsData,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    groupby = caption
  ),
  groupAesthetics = c()
)

Time profile plot showing multiple simulations and observed data without legend. By setting groupAesthetics to empty, both simulated and observed data are displayed without color grouping or legend entries, creating a simplified visualization focusing on the data patterns.

2.3.5 Observed Data with Shape as Gender

In this example, observed data is used as both simulated and observed data, connecting the different data points with a thin line.

plotTimeProfile(
  data = obsDataGender,
  observedData = obsDataGender,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    groupby = caption,
    shape = gender
  ),
  geomLineAttributes = list(linetype = "solid", linewidth = 0.5)
) +
  theme(legend.position = "right")

Time profile plot using observed data as both simulated and observed data layers. Shows how observed data points can be connected with thin lines and differentiated by shapes representing gender, creating a hybrid visualization with both line connections and scatter points.

A similar plot can be produced by combining caption and gender with interaction.

plotTimeProfile(
  data = obsDataGender,
  observedData = obsDataGender,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    groupby = interaction(caption, gender)
  ),
  geomLineAttributes = list(linetype = "solid", linewidth = 0.5)
) +
  theme(legend.position = "right")

Time profile plot using interaction of caption and gender for grouping. Similar to previous plot but combines caption and gender into single interaction groups, creating distinct line and point combinations for each unique caption-gender combination with unified legend entries.

2.3.6 Data with Secondary Axis

In this example, a plot is generated with concentration on the left y-axis and fraction on the right y-axis. First, the metaData variable has to be adjusted. For the primary y-axis, the column “values” is mapped, and metaData provides the dimension “Concentration” for this column. A new entry “y2” is added for the secondary y-axis.

metaDataY2 <- list(
  time = list(dimension = "Time", unit = "h"),
  values = list(dimension = "Concentration", unit = "mg/l"),
  y2 = list(dimension = "Fraction", unit = "")
)

The mapping y2axis must be logical. In this example, it is (dimension == "fraction"). For the primary y-axis (concentration), a log scale is displayed, and for the secondary (fraction), a linear scale is used. The limits of the secondary axis are set to [0, 1].

plotTimeProfile(
  data = simData2Dimension,
  observedData = obsData2Dimension,
  mapping = aes(
    x = time,
    y = values,
    error_relative = error_relative,
    lloq = lloq,
    y2axis = (dimension == "fraction"),
    groupby = dimension,
    shape = caption
  ),
  metaData = metaDataY2,
  yscale = "log",
  yscale.args = list(limits = c(0.01, NA)),
  y2scale = "linear",
  y2scale.args = list(limits = c(0, 1))
) +
  theme(
    axis.title.y.right = element_text(angle = 90),
    legend.position = "right",
    legend.box = "vertical"
  )

$Time profile plot with dual y-axes showing concentration and fraction data. The left y-axis displays concentration values on log scale, while the right y-axis shows fraction values on linear scale. Data includes both simulated lines and observed points with error bars and LLOQ handling.$

3. Plot Configuration

3.1 Example for Changing Geom Attributes

The plot from section 2.3.2 was adjusted using geom attributes:

geomLineAttributes = list(linetype = 'solid'): The lines used for the simulated data are set to solid in both datasets. Note that the line type for the error bars and ribbon edges remains unchanged.
geomErrorbarAttributes = list(): The default settings for geomErrorbarAttributes, width = 0, were removed so that the bar caps are now visible.
geomRibbonAttributes = list(alpha = 0.1): The shade of the ribbons was decreased by setting the alpha to 0.1; the default value for color = NA was omitted, making the edges visible.
geomPointAttributes = list(size = 7): The size of the symbols was increased.

mapSimulatedAndObserved <- data.frame(
  simulated = unique(simData$caption),
  observed = rev(unique(obsData$caption))
)

plotTimeProfile(
  data = simData,
  observedData = obsData,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    ymin = minValues,
    ymax = maxValues,
    groupby = caption
  ),
  geomLineAttributes = list(linetype = "solid"),
  geomErrorbarAttributes = list(width = 3),
  geomRibbonAttributes = list(alpha = 0.1),
  geomPointAttributes = list(size = 7),
  mapSimulatedAndObserved = mapSimulatedAndObserved
)

Time profile plot demonstrating customized geometric attributes. Shows modified line types (solid), larger error bar width, reduced ribbon transparency (alpha=0.1), and increased point size (size=7), illustrating how to customize the visual appearance of plot elements.

3.2 Example for Changing Color Scales

3.2.1 Without Mapping Table

For plots showing only simulated or observed data, or plots with a common legend (see section 2.3.1), the colors are changed using ggplot2 functions like scale_color_manual.

Below, the plot from section 2.3.1 is repeated.

# Create datasets with common caption
simData <- data.frame(simData) %>%
  dplyr::mutate(captionCommon = gsub("Simulated ", "", caption))

obsData <- data.frame(obsData) %>%
  dplyr::mutate(captionCommon = gsub("Observed ", "", caption))

plotTimeProfile(
  data = simData,
  observedData = obsData,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    ymin = minValues,
    ymax = maxValues,
    groupby = captionCommon
  )
) +
  scale_color_manual(values = c("Data 1" = "darkred", "Data 2" = "darkgreen")) +
  scale_fill_manual(values = c("Data 1" = "red", "Data 2" = "green"))

Time profile plot demonstrating custom color scaling with manual color assignment. Shows simulated lines with ribbons and observed points using custom color palettes (darkred/red for Data 1, darkgreen/green for Data 2) applied via scale_color_manual and scale_fill_manual functions.

3.2.2 With Mapping Table

It is possible to add columns with aesthetics to the table used to map simulated and observed data. The column headers must correspond to one of the aesthetics defined in groupAesthetics.

In the example below, this is done for ‘color’ and ‘fill’:

# Define Data Mappings
mapSimulatedAndObserved <- data.frame(
  simulated = unique(simData$caption),
  observed = unique(obsData$caption),
  color = c("darkred", "darkgreen"),
  fill = c("red", "green")
)

knitr::kable(mapSimulatedAndObserved)

simulated	observed	color	fill
Simulated Data 1	Observed Data 1	darkred	red
Simulated Data 2	Observed Data 2	darkgreen	green


plotTimeProfile(
  data = simData,
  observedData = obsData,
  metaData = metaData,
  mapping = aes(
    x = time,
    y = values,
    ymin = minValues,
    ymax = maxValues,
    groupby = caption
  ),
  mapSimulatedAndObserved = mapSimulatedAndObserved
)

Time profile plot demonstrating custom color scaling using a mapping table for aesthetics. The plot combines simulated and observed data, with simulated data represented as colored lines and ribbons, and observed data as points with corresponding colors. Custom colors are assigned to each dataset via the mapping table, enhancing the visual differentiation between the datasets.

Changing the scales when using the observed simulation mapping table can also be done by adding scales manually, but it is a bit more complicated. If mapSimulatedAndObserved is not null, a reset of all relevant scales is done before plotting the observed data. The scales for the simulated data must be set before this reset. You have to call plotTimeProfile two times:

Call plotTimeProfile() for simulated data only.
Set scales for simulated data.
Call plotTimeProfile() for observed data only with the simulated plot as input plotObject.
Set scales for observed data.

mapSimulatedAndObserved <- data.frame(
  simulated = unique(simData$caption),
  observed = rev(unique(obsData$caption))
)

# Define Data Mappings
mapping <- aes(
  x = time,
  y = values,
  ymin = minValues,
  ymax = maxValues,
  groupby = caption
)

plotObject <- plotTimeProfile(
  data = simData,
  metaData = metaData,
  mapping = mapping,
  mapSimulatedAndObserved = mapSimulatedAndObserved
) +
  scale_color_manual(values = c("Simulated Data 1" = "darkred", "Simulated Data 2" = "darkgreen")) +
  scale_fill_manual(values = c("Simulated Data 1" = "red", "Simulated Data 2" = "green"))

plotObject <- plotTimeProfile(
  plotObject = plotObject,
  observedData = obsData,
  mapping = mapping,
  mapSimulatedAndObserved = mapSimulatedAndObserved
) +
  scale_color_manual(values = c("Observed Data 1" = "darkred", "Observed Data 2" = "darkgreen"))
#> Scale for x is already present.
#> Adding another scale for x, which will replace the existing scale.
#> Scale for y is already present.
#> Adding another scale for y, which will replace the existing scale.

plot(plotObject)

3.3 Example for Adjusting X and Y Scale

In the example, we set the scale for the y-axis to log scale. By default, a time profile plot starts at 0; here, the defaults were overwritten, and the breaks were set manually.

plotTimeProfile(
  data = simData %>% dplyr::filter(values > 0),
  metaData = metaData,
  mapping = aes(
    x = time + 24,
    y = values,
    groupby = caption
  ),
  yscale = "log",
  xscale.args = list(limits = c(24, 48), breaks = seq(24, 48, 3))
)

Time profile plot showing adjusted y-axis scale to log scale, with customized x-axis limits and breaks. The plot visualizes simulated data filtered to values greater than zero, illustrating how to manipulate axis scales effectively.

3.4 Adjust Time Unit

The breaks of the time axis are set according to the units provided by the variable metaData.

Below, we show the same plot with four different time units:

plotlist <- list()

for (unit in c("h", "day(s)", "weeks(s)", "month(s)")) {
  metaData$time$unit <- unit
  plotlist[[unit]] <- plotTimeProfile(
    data = simData,
    metaData = metaData,
    mapping = aes(
      x = time,
      y = values,
      groupby = caption
    )
  ) + labs(title = paste("Time unit:", unit)) + theme(legend.position = "none")
}

cowplot::plot_grid(plotlist = plotlist, labels = "AUTO")

Grid of time profile plots demonstrating the effect of different time units on the x-axis. Each plot is labeled according to the time unit used, illustrating how time unit adjustments can influence the visualization of time profile data.