home

For PPI Clustering

top

General description

This document describes how to use the MACCU, which was originally designed for clustering based on array data, to do PPI clustering based on PPI (protein-protein interaction) relationships.
top

Data file correspondence

When using the MACCU, you shall need at least the following files as input:
  1. -I: a tab-delimited table file contains (a) slide names in the first row, (b) probe names in the first column, and (c) (normalized) expression levels.
  2. (optional) -C: pre-computed probe-to-probe correlation values.
  3. -assign: a mapping file that maps probe names (in above two files) to accessions like Gene IDs.
  4. -P: gene list, or probe list
When doing PPI clustering, all things are so similar, and there should be the following things:
  1. a computed/curated protein-protein interaction datafile, maybe with a confidence value for each PPI.
  2. a gene/protein list that you want to do PPI clustering.
For example, to do PPI clustering based on Interactome2.0 using MACCU, we need the following two files:
  1. DummyExp.txt: which contains dummy expression information because the expression file (assigned by -I option) will no longer be useful if there is a pre-computed correlation file (assigned by -C option)
  2. DummyMapping.txt: which contains dummy Gene-ID-to-Gene-ID mapping because there is no probe in PPI clustering analysis
So for doing PPI clustering using MACCU, just remember to use options as follows:
  1. -I DummyExp.txt
  2. -C Interactome.correl
  3. -assign DummyMapping.txt
  4. -P: your gene list
Also note that, in the Interactome2.0, the confidence values of PPIs are all above or equal to 1. Just remember to set the -CF (correlation filter) to be 1, 2, or 11 for low-, medium-, or high-confidence, respectively.