Evaluation of high-throughput functional categorization of human disease genes-0

<p><b>Copyright information:</b></p><p>Taken from "Evaluation of high-throughput functional categorization of human disease genes"</p><p>http://www.biomedcentral.com/1471-2105/8/S3/S7</p><p>BMC Bioinformatics 2007;8(Suppl 3):S7-S7.</p><p>Published online 9 May 2007</p><p>PMCID:PMC1892104.</p><p></p>rrespond to the overall accuracy scores of the two methods. Additional points on the precision-recall curve were obtained by progressively removing classes with poor precision from the evaluated set. Vale's Human Disease Gene annotations was used as a Gold Standard to calculate precision and recall, the task being to recapitulate Valle's categorization of human disease genes via GO Mapping or GO Clustering as described in the methods. Note that between 30%–55% recall, the GO Clustering Method provides higher precision than the GO Mapping. Overall, GO Mapping provides about 15% higher recall, but 10% lower precision than GO Clustering