Captor

Captor predicts O-glycosylation sites in Homo sapiens proteins to identify positions of O-linked posttranslational modifications.

Key Features:

• Homo sapiens dataset: Compiled a curated dataset specific to Homo sapiens for model training and evaluation.
• Imbalanced data handling: Employs random undersampling and the synthetic minority oversampling technique (SMOTE) to balance class distributions.
• Feature optimization: Uses the Kruskal-Wallis (K-W) test to refine feature vectors and retain informative attributes.
• Classifier: Trains a support vector machine (SVM) selected after comparison with other traditional machine learning classifiers and deep learning models.
• Comparative performance: Evaluated on independent test sets and demonstrated improved predictive capability over existing O-glycosylation prediction tools.

Scientific Applications:

• O-glycosylation site annotation: Predicts O-linked glycosylation positions in human proteins for annotation of posttranslational modifications.
• Experimental design and interpretation: Guides selection and interpretation of experiments targeting O-glycosylation in Homo sapiens proteins.
• Study of physiological and pathological processes: Supports investigation of roles of O-glycosylation in human biology and disease mechanisms.

Methodology:

Compiled a Homo sapiens dataset; applied random undersampling and SMOTE for class balancing; used the Kruskal-Wallis (K-W) test for feature refinement; trained and optimized a support vector machine (SVM) after comparison with other classifiers and deep learning models; evaluated on independent test sets.