SSKM Succ

SSKM Succ predicts succinylation sites on lysine residues in proteins to identify positions of this post-translational modification (PTM).

Key Features:

• Semi-supervised negative-sample identification: Implements a novel semi-supervised learning method to identify reliable non-succinylation lysine sites for model training.
• K-means clustering: Applies K-means clustering to partition data into five distinct clusters.
• Sequence feature representations: Utilizes grey pseudo amino acid composition, K-space, and position-special amino acid propensity to represent protein sequences.
• Two-step feature selection: Employs a two-step feature selection process to remove redundant features and optimize models for each cluster.
• Support vector machine models: Uses support vector machine (SVM) classifiers to construct individual prediction models tailored to each cluster.
• Performance: Achieves 80.18% accuracy on independent testing datasets as reported.

Scientific Applications:

• PTM site identification: Enables prediction of protein succinylation sites to support studies of lysine-specific post-translational modification.
• Pathway analysis: Facilitates investigation of succinylated proteins in amino acid degradation and fatty acid metabolism pathways.
• Disease association studies: Supports exploratory analyses of potential links between protein succinylation and neurodegenerative diseases.

Methodology:

Employs a novel semi-supervised learning method; partitions data via K-means clustering into five clusters; represents sequences with grey pseudo amino acid composition, K-space, and position-special amino acid propensity; applies a two-step feature selection process to eliminate redundant features; and builds cluster-specific predictors using support vector machine (SVM).