Cooperative Co-evolutionary Module Identification with Application to Cancer Disease Module Discovery

Abstract

Module identification or community detection in complex networks has become increasingly important in many scientific fields because it provides insight into the relationship and interaction between network function and topology. In recent years, module identification algorithms based on stochastic optimization algorithms such as Evolutionary Algorithms have been demonstrated to be superior to other algorithms on small to medium scale networks. However, the scalability and resolution limit problems of these module identification algorithms have not been fully addressed, which impeded their application to real-world networks. This paper proposes a novel module identification algorithm called Cooperative Co-evolutionary Module Identification to address these two problems. The proposed algorithm employs a cooperative co-evolutionary framework to handle large scale networks. We also incorporate a recursive partitioning scheme into the algorithm to effectively address the resolution limit problem. The performance of our algorithm is evaluated on twelve benchmark complex networks. As a medical application, we apply our algorithm to identify disease modules that differentiate low and high grade glioma tumours to gain insights into the molecular mechanisms that underpin the progression of glioma. Experimental results show that the proposed algorithm has a very competitive performance compared with other state-of-the-art module identification algorithms.

Discovering Essential Multiple Gene Effects Through Large Scale Optimization: An Application to Human Cancer Metabolism

“…Genetic algorithms are advanced heuristic methods that have been successfully applied to solve hard problems involving biological optimization of given objective functions in a wide range of parameter selection and control problems. These include knockout-or partial knockdown-based metabolic engineering [28], robust systems design [29], evaluation of response to cancer treatment [30], identification of cancer glioma tumors [31], or design of chemotherapies [32].…”

Constructing Real-Life Benchmarks for Community Detection by Rewiring Edges

“…However, the robustness of algorithms can hardly be tested on real-life benchmarks directly because we do not know exactly how well the algorithms will perform if the community intensity of the original network changes. erefore, in most previous studies, evaluating the robustness of a community detection algorithm can only be carried out on artificial synthetic benchmarks (e.g., GN and LFR) [8][9][10][11][12][13][28][29][30][31], whose structural characteristics differ significantly from real-life networks [14,15,[23][24][25][26][27].…”

“…As the number of each real-life network is only one and the community intensity within the network cannot be adjusted, it is impossible to know how the detection accuracy of the algorithm changes when the network community structure gradually blurs. erefore, in the existing research studies, the robustness evaluation of algorithms still needs to rely on synthetic benchmarks, such as GN and LFR, which can be constructed as a series of networks with gradually varied community intensity [12,13,[28][29][30][31]. However, the modification of mesoscale structures (i.e., community) in existing synthetic benchmarks, often changes with microscale characteristics, which affects the accuracy of robustness assessment.…”

Active module identification in intracellular networks using a memetic algorithm with a new binary decoding scheme

“…In our previous works, we have shown that by incorporating local search operators into generic metaheuristic optimization algorithms, we can significantly improve the speed and accuracy for community detection in large scale biological networks [16, 17]. …”

“…Another problem of using generic metaheuristic optimization algorithms is that the search operators, i.e., perturbation [ 5 ], mutation and crossover [ 14 ], are not specifically designed for active module identification, which might result in mediocre search performance in terms of speed and accuracy. In our previous works, we have shown that by incorporating local search operators into generic metaheuristic optimization algorithms, we can significantly improve the speed and accuracy for community detection in large scale biological networks [ 16 , 17 ].…”