Application Note in BIOINFORMATICS


Identifying Significant Co-interaction Patterns

SICOP is the implementation of a widely applicable method to data sets that describe one or two types of connections between elements of two different types (bipartite networks) such as cellular variables under the influence of different experimental conditions, agents affiliated with societies, customers buying, renting or rating products, or authors writing scientific papers. It evaluates the statistical significance of the number of common neigbors of two elements of the same type (e.g. cellular variables, societies, products, and articles) by comparison with their expected number of common neighbors in a suitable null model. The null model consists of a sample from the family of random graphs in which the total number of nodes and edges as well as the number of connections per node is fixed.

SICOP is available under the terms of the GNU General Public Licencec. You can download it here. The package contains an executable jar file, a user's manual, and the source code.
The software is written in Java. It runs without installation or compilation on any platform with an existing Java Runtime Environment.

About the method
The theoretical background of the algorithm behind SICOP and a formal proof of the shortcomings of the more commonly used approach that relies on a simplified null model can be found in KA Zweig and M Kaufmann: A systematic approach to the one-mode projection of bipartite graphs, Social Network Analysis and Mining 1, pp. 187-218 (2011). There, the method is applied to a large-scale data set from the field of market basket analysis (the Netflix Prize data set) and the results are compared with a real-world ground truth.

Projects using SICOP
The method underlying SICOP has already been shown to provide new biological insights in the context of miRNA regulation of proteins in a human breast cancer line. The results can be found in S Uhlmann et al.: Global microRNA level regulation of EGFR-driven cell-cycle protein network in breast cancer, Molecular Systems Biology 8:569 (2012). The importance of the method for the microRNA study was emphasized in the associated News and Views article M Malumbres: miRNAs versus oncogenes: the power of social networking, Molecular Systems Biology 8: 569 (2012).
Please let us know if you are using SICOP for your research so that we can extend this list of projects!

Data sets
The following data sets have already been analysed with SICOP:

  • microRNA regulation of the EGFR-driven cell-cycle protein network in breast cancer: available as gml file here
  • MovieLens: film rating data from GroupLens Research (as txt file)

We are happy to link your data if it can be analysed with SICOP.