Prediction of mitochondrial proteins is one of the major challenge in the filed bioinformatics due to their importance living organism. Mitochondrial proteins are associated with diseases like Alzheimer, Perkinson and Type II diabetes. Thus it is important to develop method for predicting mitochondrial proteins. The existing subcellular localization methods predict most of the location with high accuracy except mitochondrial protein. In order to improve accuracy of prediction of mitochndrial protein we developed a novel method Mitpred, based on presence of exclusive mitochondrial domains.
SVM models using split amino acid composition (25 N-terminal, 25 C-terminal, and remaining residues)
HMM model for searching of exclusive mitochndrial domains
Hybrid model combines SVM and HMM model
Annotation of six organisms
Recently exclusive domains have been updated
Evaluated using five-fold cross-validation and tested on blind dataset
Server allow mapping of mitochndrial domians on users sequence