This is the code that accompanies the publication "Bayesian Ontology Querying for Accurate and Noise-Tolerant Semantic Searches".
The authors of this software have filed a patent on ontology search algorithms for several applications, including the algorithm implemented in this work.
We release the source code under the ClearBSD license. Additional, we grant the usage of the patent for academic purposes that have no commercial interests. Other parties have to contact one of the authors for obtaining permission or a license.
In order to start the benchmark, just invoke the supplied makefile. You need Java, ant, and R for successfully completing the benchmark process. Look into the code for more details about the implementation and how to use it in your own research project. Remember that it also often helps to just read the supplied tests to understand how to use an API. In particular, the BOQA implementation is tested in BOQATest class.
Given a directory hpo_dir with HPO files (*_hpo.txt) and an ouput directory out_dir, and from the root of the repo:
java -Xmx16G -cp bin:jars/commons-cli-1.2.jar sonumina.boqa.BOQABenchmark -o data/hp.obo.gz -a data/new_phenotype.gz -p <hpo_dir> -d <out_dir>Code forked from public release: http://compbio.charite.de/boqa/
Berlin, June 8th, 2012 Sebastian Bauer, Sebastian Köhler, Marcel Schulz, Peter Robinson