calculate_complex_auc: Evaluate the recovery of known protein complexes

In fosterlab/CFTK: The Co-Fractionation Toolkit

Description

Assess the intrinsic quality of a CF-MS dataset by evaluating its ability to recover known protein complexes, using receiver operating characteristic (ROC) analysis. In this analysis, the correlation coefficients between every pair of proteins in the dataset are ranked, and compared to a binary outcome variable reflecting whether the two proteins are in the same complex. The area under the curve (AUC) is returned as a measure of the ability of the CF-MS data to recover known complexes. This measure ranges from 0 to 1, with 1 representing perfect recovery, and 0.5 representing random recovery.

Usage

calculate_complex_auc(pairs, adj, score_column = "cor")

Arguments

pairs	a matrix of dimensions (# of proteins) x (# of proteins), scoring every possible protein pair, in which higher values reflect more similar pairs, e.g. as returned by score_pairs. Alternatively, a data frame of candidate protein-protein interactions, with proteins in the first two columns.
adj	an adjacency matrix between all complex proteins, with intra-complex pairs as 1s and inter-complex pairs as 0s, e.g. as returned by to_adjacency_matrix
score_column	when pairs is a data frame, the column that contains the score for each protein pair

Value

the area under the receiver operating characteristic curve

fosterlab/CFTK documentation built on Jan. 19, 2021, 10:31 p.m.