Performance of modularity maximization in practical contexts

被引:627
作者
Good, Benjamin H. [1 ,2 ]
de Montjoye, Yves-Alexandre [2 ,3 ]
Clauset, Aaron [2 ]
机构
[1] Swarthmore Coll, Dept Phys, Swarthmore, PA 19081 USA
[2] Santa Fe Inst, Santa Fe, NM 87501 USA
[3] Catholic Univ Louvain, Dept Appl Math, B-1348 Louvain, Belgium
基金
美国国家科学基金会;
关键词
COMMUNITY STRUCTURE; NETWORK BIOLOGY; ORGANIZATION; MEMBERSHIP; RESOLUTION; EVOLUTION; MODELS;
D O I
10.1103/PhysRevE.81.046106
中图分类号
O35 [流体力学]; O53 [等离子体物理学];
学科分类号
070204 ; 080103 ; 080704 ;
摘要
Although widely used in practice, the behavior and accuracy of the popular module identification technique called modularity maximization is not well understood in practical contexts. Here, we present a broad characterization of its performance in such situations. First, we revisit and clarify the resolution limit phenomenon for modularity maximization. Second, we show that the modularity function Q exhibits extreme degeneracies: it typically admits an exponential number of distinct high-scoring solutions and typically lacks a clear global maximum. Third, we derive the limiting behavior of the maximum modularity Q(max) for one model of infinitely modular networks, showing that it depends strongly both on the size of the network and on the number of modules it contains. Finally, using three real-world metabolic networks as examples, we show that the degenerate solutions can fundamentally disagree on many, but not all, partition properties such as the composition of the largest modules and the distribution of module sizes. These results imply that the output of any modularity maximization procedure should be interpreted cautiously in scientific contexts. They also explain why many heuristics are often successful at finding high-scoring partitions in practice and why different heuristics can disagree on the modular structure of the same network. We conclude by discussing avenues for mitigating some of these behaviors, such as combining information from many degenerate solutions or using generative models.
引用
收藏
页数:19
相关论文
共 71 条
  • [31] From molecular to modular cell biology
    Hartwell, LH
    Hopfield, JJ
    Leibler, S
    Murray, AW
    [J]. NATURE, 1999, 402 (6761) : C47 - C52
  • [32] Bayesian approach to network modularity
    Hofman, Jake M.
    Wiggins, Chris H.
    [J]. PHYSICAL REVIEW LETTERS, 2008, 100 (25)
  • [33] Modularity density of network community divisions
    Holmstrom, Erik
    Bock, Nicolas
    Brannlund, Johan
    [J]. PHYSICA D-NONLINEAR PHENOMENA, 2009, 238 (14) : 1161 - 1167
  • [34] Currency and commodity metabolites: their identification and relation to the modularity of metabolic networks
    Huss, M.
    Holme, P.
    [J]. IET SYSTEMS BIOLOGY, 2007, 1 (05) : 280 - 285
  • [35] Robustness of community structure in networks
    Karrer, Brian
    Levina, Elizaveta
    Newman, M. E. J.
    [J]. PHYSICAL REVIEW E, 2008, 77 (04)
  • [36] Kleinberg J, 2002, ADV NEUR IN, V14, P431
  • [37] The evolution of modularity in bacterial metabolic networks
    Kreimer, Anat
    Borenstein, Elhanan
    Gophna, Uri
    Ruppin, Eytan
    [J]. PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2008, 105 (19) : 6976 - 6981
  • [38] NONMETRIC VARIETY OF LINEAR FACTOR-ANALYSIS
    KRUSKAL, JB
    SHEPARD, RN
    [J]. PSYCHOMETRIKA, 1974, 39 (02) : 123 - 157
  • [39] Limited resolution in complex network community detection with Potts model approach
    Kumpula, J. M.
    Saramaki, J.
    Kaski, K.
    Kertesz, J.
    [J]. EUROPEAN PHYSICAL JOURNAL B, 2007, 56 (01) : 41 - 45
  • [40] Lee JA, 2007, INFORM SCI STAT, P1