Confused on how to interpret CIM heatmap in rcc method

Support
1

Hello,

It seems that in rcc method, cim function computes a correlation between each variate and each original feature and multiplies the resulting matrices:

bisect = mat$variates$X[, comp] + mat$variates$Y[, comp]
cord.X = cor(mat$X, bisect, use = "pairwise")
cord.Y = cor(mat$Y, bisect, use = "pairwise")
XY.mat = as.matrix(cord.X %*% t(cord.Y))

Multiplying correlation coefficients by each other isn’t a very common operation in statistics, so I wonder if the resulting matrix has any straightforward interpretation?

Thank you.

2

hi @plezar,

Yes, we calculate the correlation with the component and then multiply it. We’ve shown in simulated data that it works well in practice.

You can find more details about why in the method section in: Visualising associations between paired ‘omics’ data sets | BioData Mining | Full Text or the book if you have access to it through your library Book | mixOmics (chapter 6 Appendix).

Kim-Anh