To study whether the predictions of the individual methods result in the same or complementary sets of predictions we determined the overlap of GO terms predicted by either strategy.
The GO terms predicted by the system are compared with gold standard to calculate precision, recall and F-measure.
Yet, the quality of the GO terms predicted by machine learning continues to improve, thanks to the growing number of high-quality GO terms assignments available in GOA: since 2006, GOCat performances have improved by ∼50%.
In addition, we assessed the performance of: c) an indirect method that first predicts disease terms for a gene using a structured model and then maps them to HPO terms and d) using OMIM disease terms predicted by PhenoPPIOrth followed by mapping the OMIM terms to HPO terms.
Since a method that detects associations between a TF and highly specific GO terms is more useful than one that reports only general GO terms, we also measure the average depth of the most specific GO terms predicted by G omo.
Moreover, we showed in (46) that the quality of the GO terms predicted by GOCat continues to improve across the time, thanks to the growing number of high-quality GO terms assignments available in GOA: thus, since 2006, GOCat performances have improved by +50%.
One of the best set of tools we have has recently been coined with the term "CleanWeb," predicted by the chair of the Clean Economy Network as the next wave of cleantech.
For a wide range of size definitions of the promoter regions, the accuracy (mean AUC50) of TF GO term associations predicted by G omo for the related species is similar to the accuracy using the key species.
In a multivariate model, knowledge of evolutionary terms was predicted by education and religious denomination.
About the medium-term evolution predicted by the models, we identified countries with declining consumption (the US, France, Germany and South Korea) and with increasing consumption (China, Russia, and Canada).
