detect_complexes: Detect significantly interacting complexes in a chromatogram...

Description Usage Arguments Value Examples

View source: R/detect_complexes.R

Description

Use a permutation testing approach to identify complexes that show a significant tendency to interact, relative to random sets of complexes of equivalent size. The function begins by calculating the Pearson correlation or Euclidean distance between all proteins in the matrix, and

Usage

1
2
3
4
5
6
7
8
detect_complexes(
  profile_matrix,
  complexes,
  method = c("pearson", "euclidean"),
  min_pairs = 10,
  bootstraps = 100,
  progress = TRUE
)

Arguments

profile_matrix

a matrix of chromatograms, with proteins in the rows and fractions in the columns, or a MSnSet object

complexes

a named list of protein complexes, where the name is the complex name and the entries are proteins within that complex

method

method to use to calculate edge weights; one of pearson or euclidean

min_pairs

the minimum number of pairwise observations to count a correlation or distance towards the z score

bootstraps

number of bootstraps to execute to estimate z scores

progress

whether to show the progress of the function

Value

a named vector of z scores for each complex in the input list

Examples

1
2
3
4
5
6
data(scott)
data(gold_standard)
complexes <- gold_standard[lengths(gold_standard) >= 3]
z_scores <- detect_complexes(t(scott), complexes)
length(na.omit(z_scores)) ## number of complexes that could be tested
z_scores[which.max(z_scores)] ## most significant complex

fosterlab/PrInCE documentation built on Dec. 13, 2020, 5:50 a.m.