calculateGeneScores: Calculate the Prix Fixe gene score

In princeew/PFFindR: Identify, evaluate, and visualize enriched Prix Fixe subnetworks

Description

Evaluate the gene contributions to a given subnetwork.

Usage

calculateGeneScores(network, pf_data)

Arguments

network	A PrixFixeNetwork. Probably selected using getTopNetwork.
pf_data	A PFData object.

Scoring Methodology

To score a gene contribution to a subnetwork we first calculate the null locus scenario in which we remove, or zero out, a given locus and calculate the density of the resulting network. Next, we iteratively consider all genes associated with that locus by substituting and caculating the density. The difference between each gene density and the empty locus density is the score for that gene. If the gene has no connectivity to the subnetwork, then we give it an NA value.

To implement this method we simplify the problem. First, we are only considering one locus at a time and therefor all other loci remain constant and can be ignored here. Due to this, the null locus density must be zero because there is no connectivity by definition. Therefore, the only value we need to derive is the sum of the connections between the permuted gene and the other fixed loci.

Moreover, to quickly calculate the contributions we create a copy of a network and modify the genes to include the network genes and all genes associated with a given locus. The adjacency matrix is then recalculated, and filtered such that the rows only contain the locus genes and the columns are the network genes. The contribution for each one of the locus genes is then calculated as the corresponding row sum.

princeew/PFFindR documentation built on Dec. 31, 2020, 2:06 a.m.