Finding frequent subgraphs in biological networks via maximal item sets

We describe an improvement of an algorithm for detecting frequently occurring patterns and modules in biological networks. The improvement is based on the observation that the problem of finding frequent network parts can be reduced to the problem of finding maximal frequent item sets (MFI). The MFI problem is a classical problem in the data mining community and there exist numerous efficient tools for it, most of them publicly available. We apply MFI tools to find frequent subgraphs in metabolic pathways from the KEGG database. Our experimental results show that, compared to the existing specialized tools for frequent subgraphs detection, the MFI tools coupled with an adequate postprocessing are much more efficient with regard to the running time and the size of the frequent graphs