Mal-Prec

Mal-Prec predicts malonylated sites in protein sequences to identify lysine malonylation relevant to studies of post-translational modification and disease associations such as Type 2 Diabetes Mellitus and cancer.

Key Features:

• Machine Learning Framework: Mal-Prec employs a supervised Support Vector Machine (SVM) classifier to predict malonylation sites.
• One-hot Encoding: Uses one-hot encoding to represent amino acid identities as sequence features.
• Physicochemical Properties (AAindex): Incorporates AAindex-derived physicochemical properties of amino acids as feature inputs.
• Composition of k-spaced Amino Acid Pairs (CKSAAP): Utilizes CKSAAP to capture local sequence context and spacing of residue pairs.
• Feature Selection (PCA): Applies Principal Component Analysis (PCA) to reduce dimensionality and select optimal feature subsets.
• Validation and Performance: Evaluated using five-fold cross-validation and independent test datasets with reported AUCs of 96.47% and 90.72%, respectively, and reported to outperform existing prediction tools.

Scientific Applications:

• Identification of Malonylation Sites: Enables prediction and discovery of novel protein malonylation sites.
• Biomarker and Therapeutic Target Discovery: Supports identification of potential biomarkers and therapeutic targets related to malonylation in diseases such as Type 2 Diabetes Mellitus and cancer.
• Post-translational Modification Research: Facilitates studies in bioinformatics and molecular biology focused on post-translational modifications.

Methodology:

Protein sequences are encoded using one-hot encoding, AAindex physicochemical properties, and CKSAAP; PCA is applied for feature optimization; an SVM classifier is trained and evaluated with five-fold cross-validation and independent test datasets.