IBM Frequent Subgraph Miner

IBM Frequent Subgraph Miner is a data-mining tool that uses the a priori algorithm for finding related sets of items in transactional data and expands the existing algorithm by graphing structured data. An alternative way to model objects in a data set is to use graphs. In this approach, a labeled graph is used, constructed so that each vertex and edge are independently labeled. When the data set (consisting of a set of labeled graphs and the minimum support) is given as input, this tool derives all frequently-connected subgraphs whose support is more than or equal to the minimum support value in the data set. IBM Frequent Subgraph Miner can efficiently find connected common substructures that frequently occur in a set of labeled graphs. Graphs are especially useful in databases of chemical structures in chemical compounds. If a set of compounds is given as input, this tool can efficiently find common substructures (subgraphs). The discovered patterns from a set of compounds with a certain activity can then be used to design new drugs. This technology includes a sample data set, which consists of 340 chemical compounds; a resulting research report, which describes the algorithm used in detail; and complete instructions, which describe how to use this tool.