How To Use CORREP to Estimate Multivariate Correlation and Statistical Inference Procedures

OMICS data are increasingly available to biomedical researchers, and (biological) replications are more and more affordable for gene microarray experiments or proteomics experiments. The functional relationship between a pair of genes or proteins are often inferred by calculating correlation coefficient between their expression profiles. Classical correlation estimation techniques, such as Pearson correlation coefficient, do not explicitly take replicated data into account. As a result, biological replicates are often averaged before correlations are calculated. The averaging is not justified if there is poor concordance between samples and the variance in each sample is not similar. Based on our recently proposed multivariate correlation estimator, CORREP implements functions for estimating multivariate correlation for replicated OMICS data and statistical inference procedures. In this vignette I demo an non-trivial task accomplished using CORREP. First let’s look at examples of replicated OMICS data and non-replicated OMICS data. x11, x12, . . . , x1n