TISHunter is a highly accurate predictor for translation initiation sites in human mRNAs. TISHunter is based on support vector machines with the edit kernel. For details, please see "A Class of Edit Kernels for SVMs to Predict Translation Initiation Sites in Eukaryotic mRNAs" (in RECOMB04). Oringinally, the algorithm was tested on Pedersen and Nielsen's dataset of 3312 sequences from vertebrates and achieved 99.9% accuracy. Currently, TISHunter is trained on a collection of 8225 human mRNA sequences and the three-fold cross validation accuracy is 96.7%. Besides, the test on 5651 mouse sequences still shows a reasonably high accuracy (89.5%).
|
Please send comments and questions to Haifeng Li |
|