Skip to contents

Generate reciprocal best hits based on correlations

rs_rbh_cor.Rd

[Experimental] This function takes list of (named) matrices which represent for example matrix factorisation results you wish to identify (k-th) reciprocal best hits (RBH) for. The rows need to represent the features and the columns the parts you wish to calculate the RBH for.

Usage

rs_rbh_cor(module_matrices, k_best, spearman, min_similarity)

Arguments

module_matrices

A named list of matrices with row and column names. Rows represent features and columns the modules you wish to calculate the correlations for. Only features shared between two matrices are used.

k_best

Integer. Number of best neighbours to consider. If set to 1L, this behaves as the traditional reciprocal best hit. If you set this to 3L you consider edges if the modules is in the top 3 best modules by similarity for each other.

spearman

Boolean. Shall Spearman correlation be used.

min_similarity

Numeric. Only hits with an absolute correlation strictly above this are returned.

Value

A list containing:

  • origin - The name of the origin of the gene modules.

  • target - The name of the target of the gene modules.

  • comparisons - Integer vector indicating how many RBH hits were identified in this comparison

  • origin_modules - Names of the gene modules from the origin.

  • target_modules - Names of the gene modules from the target.

  • similarity - The absolute correlations between the two respective gene modules.