A training algorithm for optimal margin classifiers
COLT '92 Proceedings of the fifth annual workshop on Computational learning theory
Machine Learning
X-means: Extending K-means with Efficient Estimation of the Number of Clusters
ICML '00 Proceedings of the Seventeenth International Conference on Machine Learning
Sparse bayesian learning and the relevance vector machine
The Journal of Machine Learning Research
Prediction of phosphorylation sites using SVMs
Bioinformatics
Bioinformatics
Sparse Kernel Learning and the Relevance Units Machine
PAKDD '09 Proceedings of the 13th Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining
Hi-index | 0.00 |
Phosphorylation is an important post-translational modification of proteins that is essential to the regulation of many cellular process. The in vivo and in vitro discovery of phos-phorylation sites is an expensive, time-consuming and laborious task. In this preliminary study, we assess the viability of using our proposed probabilistic Classification Relevance Units Machine (CRUM) for in silico phosphorylation site prediction. We conduct a comparison with the popular Support Vector Machine (SVM) and the Relevance Vector Machine (RVM) that, unlike the SVM, has not been applied to phosphorylation site prediction. The resulting CRUM and RVM predictors offer comparable predictive performance to the SVM. The main advantages of CRUM and RVM over the SVM are: 1. An estimation of the posterior probability of the site being phosphorylatable, providing biologists an important measurement of the uncertainty of the prediction. 2. A more parsimonious model, leading to a reduction in prediction run-time that is important in predictions on large-scale data. Furthermore, the CRUM training algorithm has lower runtime and memory complexity and has a simpler parameter selection scheme than the RVM learning algorithm. Therefore we conclude that the CRUM is the most viable kernel machine for probabilistic prediction of protein phosphorylation sites.