This function generates stratified bootstrap samples similar to bootstrapSamples_limRots, but additionally supports correlation blocks. When correlation_block is specified, all samples sharing the same block ID are always selected together during resampling.

bootstrapSamples_limRots_cox(niter, meta.info, correlation_block = NULL)

Arguments

niter

Integer. The number of bootstrap samples to generate.

meta.info

Data frame. Metadata containing sample information, where each row corresponds to a sample. Factor columns in meta.info are used to define strata for sampling.

correlation_block

Character or NULL. The name of a column in meta.info that defines correlation blocks. Samples sharing the same value in this column are always resampled together as a unit.

Value

A matrix of dimension niter x n, where n is the number of samples. Each row corresponds to a bootstrap sample, and each entry is a resampled row name from the metadata.

Details

When correlation_block is not NULL, the function groups samples by their block ID within each group/stratum and resamples entire blocks with replacement, so that correlated samples (e.g., repeated measures from the same subject) are always kept together.