PredictioR is an R package for biomarker discovery and predictive modeling in immuno-oncology (IO) therapy response. It supports pan-cancer, cancer-specific, and treatment-specific analyses, and integrates clinical covariates like age, sex, and tumor type into its modeling workflows.
The package includes:
- Signature scoring methods for curated IO gene signatures
- IO response prediction algorithms
- Functions for clinical association analysis
- Built-in support for curated datasets and integration with the SignatureSets package
Data Resources
- IO Datasets: Clinical and molecular profiles of IO-treated cohorts, available at: ORCESTRA
- IO Signatures: IO gene signatures available from the companion repository: SignatureSets GitHub repository
Dependencies
Requires R 4.4.1 or higher
Install required packages:
if (!requireNamespace("BiocManager", quietly = TRUE)) install.packages("BiocManager")
dependencies <- c(
"MultiAssayExperiment", "survival", "survcomp",
"GSVA", "meta", "ggplot2", "ggrepel"
)
for (pkg in dependencies) {
if (!requireNamespace(pkg, quietly = TRUE)) {
BiocManager::install(pkg, update = FALSE)
}
}
Install PredictioR from GitHub
devtools::install_github("bhklab/PredictioR")
library(PredictioR)
For source-level exploration:
git clone https://github.com/bhklab/PredictioR
cd PredictioR
More details about function usage and computational methods are provided in the package documentation and vignettes, or via the web application at predictio.ca.
PredictioR/
├── 📁 R/ – Core package functions
├── 📁 data/ – Selected and curated IO signatures and datasets
├── 📁 man/ – Function documentation (.Rd files)
├── 📁 vignettes/ – Workflows and usage examples
├── 📄 DESCRIPTION – Package metadata
└── 📄 README.md – Overview and setup instructions
If you use PredictioR or its datasets in your work, please cite the following papers: