Background: Cluster analysis is a core task in modern data-centric computation. Algorithmic choice is driven by factors such as data size and heterogeneity, the similarity measures employed, and the type of clusters sought. Familiarity and mere preference often play a significant role as well. Comparisons between clustering algorithms tend to focus on cluster quality. Such comparisons are complicated by the fact that algorithms often have multiple settings that can affect the clusters produced. Such a setting may represent, for example, a preset variable, a parameter of interest, or various sorts of initial assignments. A question of interest then is this: to what degree do the clusters produced vary as setting values change? Results: This work introduces a new metric, termed simply “robustness”, designed to answer that question. Robustness is an easily-interpretable measure of the propensity of a clustering algorithm to maintain output coherence over a range of settings. The robustness of eleven popular clustering algorithms is evaluated over some two dozen publicly available mRNA expression microarray datasets. Given their straightforwardness and predictability, hierarchical methods generally exhibited the highest robustness on most datasets. Of the more complex strategies, the paraclique algorithm yielded consistently higher robustness than other algorithms tested, approaching and even surpassing hierarchical methods on several datasets. Other techniques exhibited mixed robustness, with no clear distinction between them. Conclusions: Robustness provides a simple and intuitive measure of the stability and predictability of a clustering algorithm. It can be a useful tool to aid both in algorithm selection and in deciding how much effort to devote to parameter tuning.

Differential Shannon entropy (DSE) and differential coefficient of variation (DCV) are effective metrics for the study of gene expression data. They can serve to augment differential expression (DE), and be applied in numerous settings whenever one seeks to measure differences in variability rather than mere differences in magnitude. A general purpose, easily accessible tool for DSE and DCV would help make these two metrics available to data scientists. Automated p value computations would additionally be useful, and are often easier to interpret than raw test statistic values alone. EntropyExplorer is an R package for calculating DSE, DCV and DE. It also computes corresponding p values for each metric. All features are available through a single R function call. Based on extensive investigations in the literature, the Fligner-Killeen test was chosen to compute DCV p values. No standard method was found to be appropriate for DSE, and so permutation testing is used to calculate DSE p values. EntropyExplorer provides a convenient resource for calculating DSE, DCV, DE and associated p values. The package, along with its source code and reference manual, are freely available from the CRAN public repository at http://cran.r-project.org/web/packages/EntropyExplorer/index.html .

Let k denote an integer greater than 2, let G denote a k-partite graph, and let S denote the set of all maximal k-partite cliques in G. Several open questions concerning the computation of S are resolved. A straightforward and highly-scalable modification to the classic recursive backtracking approach of Bron and Kerbosch is first described and shown to run in O(3^n/3) time. A series of novel graph constructions is then used to prove that this bound is best possible in the sense that it matches an asymptotically tight upper limit on |S|. The task of identifying a vertex-maximum element of S is also considered and, in contrast with the k = 2 case, shown to be NP-hard for every k ≥ 3. A special class of k-partite graphs that arises in the context of functional genomics and other problem domains is studied as well and shown to be more readily solvable via a polynomial-time transformation to bipartite graphs. Applications, limitations, potentials for faster methods, heuristic approaches, and alternate formulations are also addressed.

Objective: Bipartite graphs are widely used to model relationships between pairs of heterogeneous data types. Maximal bicliques are foundational structures in such graphs, and their enumeration is an important task in systems biology, epidemiology and many other problem domains. Thus, there is a need for an efficient, general purpose, publicly available tool to enumerate maximal bicliques in bipartite graphs. The statistical programming language R is a logical choice for such a tool, but until now no R package has existed for this purpose. Our objective is to provide such a package, so that the research community can more easily perform this computationally demanding task. Results: Biclique is an R package that takes as input a bipartite graph and produces a listing of all maximal bicliques in this graph. Input and output formats are straightforward, with examples provided both in this paper and in the package documentation. Biclique employs a state-of-the-art algorithm previously developed for basic research in functional genomics. This package, along with its source code and reference manual, are freely available from the CRAN public repository at https://cran.r-project.org/web/packages/biclique/index.html.