Composite measures schemes (cms)
The cms package uses tabular data from rat epilepsy studies and applies a composite measures scheme (via PCA) to select the most prominent features. Further, variables can be selected to perform cluster analysis on a subset in order to build a composite score. Finally, the cluster distribution is displayed for the subgroups and allows severity assessment between animal models.
Please note: the cms_analysis and cms_cluster functions are deprecated.
Click here for reading the cms Vignette.
Dependencies
The cms package has some dependencies. We advise installing/updating the following packages before using cms:
- ggplot2
- dplyr
- factoextra
Installation
You can install the development version from GitHub with:
# install.packages("devtools")
devtools::install_github("mytalbot/cms")
library(cms)Example cms
The following example uses the (pre-cleaned) internalized epilepsy data (episet_full) set with three experimental subgroups. Further, the feature selection is repeated 100-fold. The example uses the new cms function. Please note that all variables that shall be included must be specified in the vars object.
Working example
Note: the following example shows the pooled data from the episet_full set, using the pooled subgroups. You might need to filter them, if you are interested in specific subsets.
library(cms)
# Do the cms feature analysis (with a limited set of variables)
usecase <- cms(raw = episet_full,
runs = 100,
idvariable = "animal_id",
setsize = 0.8,
variables = c("Sacc_pref", "social_interaction",
"burrowing_rat", "openfield_rat"),
maxPC = 1:4,
clusters = 3,
showplot = FALSE)
# This also shows the plot
usecase$p
Table of the cms feature frequency distributions
head(usecase$FRQ)
#> position x freq perc
#> 1 1 burrowing_rat 81 81
#> 2 1 openfield_rat 1 1
#> 3 1 Sacc_pref 11 11
#> 4 1 social_interaction 7 7
#> 5 2 burrowing_rat 7 7
#> 6 2 openfield_rat 8 8