Terminus

Terminus predicts initiator methionine cleavage (IMC) and Nα-terminal acetylation (N-Ac) in eukaryotic proteins to support analysis of protein maturation and regulation by post-translational modifications.

Key Features:

• Pattern Discovery and Decision Trees: Integrates manually detected sequence patterns with decision tree machine learning to generate classifiers.
• C4.5 Algorithm Coupled with Genetic Algorithms: Constructs decision trees using the C4.5 algorithm enhanced by genetic algorithms to produce interpretable "white-box" classifiers.
• Performance Metrics: Reports cross-validated Matthews correlation coefficients of 0.83 for IMC and 0.65 for N-Ac on a dataset of eukaryotic proteins.
• Comparative Superiority: Outperforms state-of-the-art models when predicting substrates of N-terminal acetyltransferase B (NatB) and N-terminal acetyltransferase C (NatC).
• Biological Insight Extraction: Extracts experimentally known facts from Homo sapiens IMC data without prior knowledge, enabling biological interpretation.

Scientific Applications:

• PTM Prediction and Substrate Identification: Predicts IMC and N-Ac to identify potential substrates of acetyltransferases such as NatB and NatC.
• Protein Maturation and Function Studies: Supports investigations into protein maturation, stability, and regulation mediated by N-terminal modifications.

Methodology:

Combines manually detected pattern discovery with decision tree algorithms, builds classifiers using the C4.5 algorithm enhanced by genetic algorithms, and evaluates performance by cross-validation reporting Matthews correlation coefficients.