PredHydroxy

PredHydroxy predicts proline and lysine hydroxylation sites in proteins to enable annotation and analysis of these post-translational modifications.

Key Features:

• Prediction targets: Predicts proline (hydroxyproline) and lysine (hydroxylysine) hydroxylation sites in protein sequences.
• Automated prediction: Automates the identification of potential hydroxylation sites to support large-scale proteomic analyses.
• Feature encoding: Encodes sequence information using position weight amino acid composition combined with eight high-quality amino acid indices.
• Machine learning model: Employs support vector machines (SVMs) to classify hydroxylated versus non-hydroxylated residues.
• Validation and performance metrics: Reports performance evaluated by jackknife cross-validation with AUC of 82.72% and MCC of 69.03% for hydroxyproline, and AUC of 87.41% and MCC of 66.68% for hydroxylysine.
• Structural context findings: Identifies location-specific differences between hydroxylated and non-hydroxylated residues, notably in alpha helices and turns.

Scientific Applications:

• Protein function annotation: Facilitates annotation of proteins by predicting sites of proline and lysine hydroxylation.
• Pathophysiological insights: Supports investigation of pathological processes associated with altered protein hydroxylation.
• Drug design: Provides site-specific hydroxylation information that can inform drug design strategies targeting diseases related to abnormal hydroxylation.

Methodology:

Uses position weight amino acid composition, eight high-quality amino acid indices, support vector machines (SVMs), and jackknife cross-validation.