Intelligent Exploration of Negative Interaction from Protein-Protein Interaction Network and its Application in Healthcare

Abstract

Predicting protein-protein interactions (PPIs) has attracted much attention in recent years. Complexity of living systems depends on these interactions, as it controls healthy and diseased states in any organism.  Even though recent advances in high throughput technologies have amplified PPI data repository; high level of noise, sparseness and skewed degree distribution of data has been a hindrance in making any useful findings from these data. Most of the works in this area concentrated on missing link prediction, and only very few explored the possibility of predicting negative links, or links that might disappear from the network. This paper proposes a method to predict these negative links from PPI network using an adaptive genetic algorithm, which is further optimized using Minimum Weak Edge-Edge Domination (WEED) set.  The promising result obtained on MINT dataset asserts that the method can improve the quality of PPI data.