Local search for constrained graph clustering in biological networks

Semi-supervised or constrained graph clustering incorporates prior information in order to improve results. Pairwise constraints are often utilized guide the process. This work addresses a problem biological networks where (1) subgraph connectivity strictly required be satisfied and (2) quality is assessed with respect pairwise constraint violations. Existing methods fail fully satisfy constraints. paper presents an iterated local search algorithm which aims find highest short computing time. Experiments demonstrate how proposed solutions of good quality, being optimal. Additionally, method significantly outperforms existing branch-and-cut terms computational runtime produces competitive results regard other techniques algorithms. Furthermore, multilevel for designed handle large-scale graphs. The performance overall scheme variety coarsening from literature studied on large number exceeding 10,000 genes.