Post date: 30-jan-2013 12:46:34
For many species, genomes are being sequenced, but what are all the genes in those genomes doing? To address that question, ultimately experiments will be needed, but to guide these, computational predictions are very helpful. Recently, a world-wide comparison of the performance of various methods for gene function prediction has been performed. The results of this comparison are now available online. Our method, BMRF, had a quite good performance (#1 for biological process prediction in human and #2 for biological process prediction in Arabidopsis thaliana). Unfortunately, as explained in the paper, we did not participate in the predictions for prokaryotes, because originally these were not supposed to be taken into account in the evaluation. Because they are included after all, our good performance is not immediately clear from the main manuscript but you can find it back by going through the supporting information.
Note that in view of the enormous amount of sequence data, for many genes no experimental evidence will in fact be available in the near future, which indicates even more the importance of gene function prediction. We are currently working on further development of our method, integration with other methods, and application to important plant species.