NeddPred

NeddPred predicts lysine neddylation sites within protein sequences to identify post-translational modification positions relevant to protein regulation and disease.

Key Features:

• Bi-profile Bayes feature extraction: Encodes potential neddylation sites using bi-profile Bayesian probabilistic models to capture sequence features distinguishing sites from non-sites.
• Fuzzy Support Vector Machine (SVM): Classifies lysine residues using a fuzzy SVM that mitigates data noise and class imbalance.
• Performance metrics: Reports predictive performance with a Matthew's correlation coefficient (MCC) of 0.7082 and an area under the ROC curve (AUC-ROC) of 0.9769.
• Comparative advantage: Independent testing shows superior prediction accuracy compared to existing tools such as NeddyPreddy.

Scientific Applications:

• Large-scale neddylation site analysis: Enables computational screening of protein datasets to identify candidate lysine neddylation sites for downstream study.
• Investigation of regulatory roles: Supports studies on how neddylation affects protein regulation and cellular processes.
• Target identification for disease research: Assists in identifying potential therapeutic targets related to diseases associated with neddylation abnormalities.

Methodology:

NeddPred applies bi-profile Bayes feature extraction to encode sequence characteristics and a fuzzy support vector machine to classify lysine residues, with the fuzzy SVM addressing noise and class imbalance.