MethylSurroGetR is an R package designed to facilitate the construction of DNA methylation surrogate biomarker values from existing studies. It provides comprehensive tools for handling missing methylation data through various imputation strategies, including mean/median imputation for missing observations and reference-based filling for missing probes. The package supports flexible surrogate calculation with “linear”, “probability” (logistic), and “count” (Poisson) transformations, comprehensive input validation, and detailed diagnostic reporting. Additional utilities include conversion between beta values and M-values, as well as missing data analysis.
Installation
You can install the most current version of MethylSurroGetR from GitHub with:
# install.packages("remotes")
remotes::install_github("jagoode27/MethylSurroGetR")Example
This is a basic example workflow:
library(MethylSurroGetR)
# Load example data
data(beta_matrix_comp)
data(wts_df)
data(ref_df)
# Create weights vector
wts_vec_lin <- with(wts_df, setNames(wt_lin, rownames(wts_df)))[!is.na(wts_df$wt_lin)]
# Create surrogate object
my_surro <- surro_set(
methyl = beta_matrix_comp,
weights = wts_vec_lin,
intercept = "Intercept"
)
# Fill missing probes with reference values
ref_vec_mean <- with(ref_df, setNames(mean, rownames(ref_df)))[!is.na(ref_df$mean)]
my_surro <- reference_fill(my_surro, reference = ref_vec_mean)
# Calculate predictions
predictions <- surro_calc(my_surro, transform = "linear")Getting Help
See the Getting Started vignette for detailed examples and workflows.
For bug reports and feature requests, please open an issue.