Protein-protein interaction (and other biological) networks are often used to learn new biological function from their topology. Since current biological networks are noisy, their computational de-noising via link prediction (LP) could improve the learning accuracy. LP uses the existing network topology to predict missing and spurious links. Many of existing LP methods rely on shared immediate neighborhoods of the nodes to be linked. As such, they have limitations. Thus, in order to comprehensively study what are the topological properties of nodes in a network that dictate whether the nodes should be linked, we introduce novel sensitive (graphlet-based) LP measures that are overcome the limitations of the existing methods and outperform them. For details, see our paper listed below.

Reference: Yuriy Hulovatyy, Ryan W. Solava, and Tijana Milenkovic. "Revealing Missing Parts of the Interactome via Link Prediction." PLOS ONE 9.3 (2014): e90073.

Poster: Our poster from ISMB 2014 is available here.

Software: Implementation of our graphlet-based link prediction methodology is available here.

Data: Networks used in our study (including randomized data) are available here.

Results: The list of all node pairs predicted as edges by any of the considered methods is available here.