Introduction
The Coxmos R package is an end-to-end pipeline designed for the study of survival analysis for high dimensional data. Updating classical methods and adding new ones based on sPLS technologies. Furthermore, includes multiblock functions to work with multiple sets of information to improve survival accuracy.
The pipeline includes three basic analysis blocks:
Computing cross-validation functions and getting the models.
Evaluating all the models to select the better one for multiple metrics.
Understanding the results in terms of the global model and the original variables.
Coxmos contains the necessary functions and documentation to obtain from raw data the final models after compare them, evaluate with test data, study the performance individually and in terms of components and graph all the results to understand which variables are more relevant for each case of study.
Installation
Dependencies requiring manual installation
Some of the metrics available in Coxmos are optional based and will not be included in the standard Coxmos installation. A list of all optional packages are shown below:
- nsROC:
- smoothROCtime:
- survivalROC:
- risksetROC:
- ggforce:
- RColorConesa:
Installing Coxmos
The Coxmos R package and all the remaining dependencies can be installed from CRAN:
install.packages("Coxmos")Or from GitHub using devtools
devtools::install_github("BiostatOmics/Coxmos")In case of using Github, to access vignettes, you will need to force building with
devtools::install_github(build_vignettes = TRUE). Please note that this will
also install all suggested packages required for vignette build and might
increase install time. Alternatively, an HTML version of the vignette is
available under the vignettes
folder.
Getting started
In order to use Coxmos, you will need the following items:
- An explanatory X matrix.
- A response survival Y matrix (with two columns, "time" and "event").
Please note that two toy datasets are included in the package. Details to load and use them can be found in the package's vignette.
Contact
If you encounter a problem, please open an issue via GitHub.
References
If you use Coxmos in your research, please cite the original publication: