TransMiner: Mining Transitive Associations among Biological Objects from Medline

Summary: Associations among biological objects such as genes, proteins, and drugs can be discovered automatically from scientific literature. This paper describes ‘TransMiner’ a system for finding interesting associations among objects by mining the Medline database of scientific literature. The direct associations among the objects are discovered based on the principle of cooccurrence from the text documents. The principle of transitive closure is applied to the association graph to find potential transitive associations. The potential transitive associations that are indeed, direct are discovered by iterative retrieval and mining of the Medline database. Those associations that are not found explicitly in the entire Medline database are transitive associations and are candidates for hypotheses generation. All the discovered direct associations were manually evaluated. The direct and transitive associations are visualized using a graph visualization applet for use by the scientists. TransMiner was tested by finding associations among 56 breast cancer genes and by finding associations among 24 objects in calpain signal transduction pathway. Out of 413 direct associations discovered by TransMiner among 56 gene symbols involved in breast cancer, 329 direct associations (79.66%) were found to have some valid biological association. Out of 159 direct associations discovered by TransMiner among 24 objects involved in calpain signal transduction pathway, 155 direct associations (97.48%) were found to have some valid biological association. Availability: Graph visualization applets and result tables are available at http://sifter.cs.iupui.edu/~sifter/transMiner