SQUEAK - Statistical Quantification & Understanding of Experimental Animal Kinetics

SQUEAK is a simple package to perform regression modelling and subsequent visualisation of mouse tumour growth data. It is designed to provide accessible access to high quality statistical analysis to support comparisons between groups, while accounting for the inherent variability between individual animals.

This package stands on the shoulders of giants, and wraps a number of excellent other tools that should be given due credit, principally: nlme, ggplot2 and sjPlot.

Input Data

SQUEAK comes with a test dataset that can be explained as follows.

If you can convert your data into a similar file format, you should be in a good spot to do this kind of analysis! SQUEAK expects the input data to be in the form of an R data.frame.

Installation

You can install the development version of SQUEAK from GitHub with:

# install.packages("devtools")
devtools::install_github("seedgeorge/SQUEAK")

Minimal Usage Example

The following code uses a dataset included in the package, performs a simple data check, and then makes a simple plot.

library(SQUEAK)
#> Loading required package: ggplot2
data("long_mice")
ready_data = check_long(long_mice,
                        timecol = "Days",
                        IDcol = 'Number',
                        groupcol = 'Group',
                        measurementcol = 'Value',
                        reference_level = 'Control')
#> ✔ All columns detected!
#> ✔ Data types good!
#> ℹ 4 Groups Detected
#> ℹ Data structure suggests nested model, with unique individuals per group.
#> ℹ Using reference level: "Control"
#> ! 294 rows with NA detected, dropping these rows.
#> ✔ 181 rows remaining.
#> Data statistics:
#> ℹ 19 Individuals
#> ℹ Start Time: 0
#> ℹ End Time: 24
#> ℹ Minimum measurement: 0.03106832625
#> ℹ Maximum measurement: 1.42182667828125
colors = set_palette(ready_data)

plot_raw_lines(ready_data = ready_data, palette = colors) 
#> ! No title specified, using default.
#> ! No subtitle specified, using default.
#> Scale for colour is already present.
#> Adding another scale for colour, which will replace the existing scale.

In this data, each line represents one mouse, and the mice are grouped in treatment arms.

The next step is to make a mixed-effect regression model, which can also be plotted.

mixed_model = mixed_effect_model(ready_data = ready_data,random_slope = T,random_intercept = T)
#>   0:     252.11960: -0.819904  1.73393 -0.505521
#>   1:     252.11960: -0.819904  1.73393 -0.505521
plot_modelled_curves(mixed_model, grouptitle = 'Treatment',palette = colors)

And we can also access the statistical output, and see the difference in growth rate of the treatment arms to the control arm.

plot_interaction_forest(mixed_model, palette= colors)