predictPPI

This prototype predicts missing protein-protein interactions (PPIs) by:

1. finding k-defective cliques of size >= q in a PPI network, and
2. treating the missing edges inside those cliques as "noisy/missing" interactions.

Requirements

• Python 3
• bin/main_edgelist.out (binary of WODC used to enumerate k-defective cliques)

Inputs

• PPI edgelist: space-separated u v pairs, zero-based integer node IDs.
• Mapping file: one protein name per line, where line i corresponds to node i.

Example mapping (line-based)

YAL001C YBR123C YDR362C ...

Usage

From the repository:

python3 predictPPI.py PPI/edgelist.txt PPI/mapping.txt 1 11 

Arguments

• ppi_edgelist: PPI network edgelist
• mapping: mapping file (line i -> protein name for node i)
• k: allowed number of missing edges per clique
• q: minimum clique size

Optional flags

• --out-dir: output directory (default: output)
• --out-cliques: output cliques file (default: output/defective_cliques.txt)
• --out-pred: output predicted edges file (default: output/predicted_missing_edges.txt)
• --bin: path to main_edgelist.out (default: ./bin/main_edgelist.out)
• --keep-tmp: keep temporary files for debugging

Outputs

• output/defective_cliques.txt
  • Each line lists protein names in a defective clique, followed by its missing edges.
  • Format: protein1 protein2 ... proteinN<TAB>missing:p1-p2,p3-p4,...
• output/predicted_missing_edges.txt
  • Each line is a predicted missing edge: proteinA proteinB

Notes

• The script cleans up temporary files by default (tmp/ and graph.txt).
• If main_edgelist.out is missing or fails, the script raises an error.