Motifs tree: a new method for predicting post-translational modifications.

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Post-translational modifications (PTMs) are crucial for protein function and maturation.
• Existing prediction models for PTMs are either inefficient, labor-intensive, or lack interpretability, hindering biological knowledge discovery.
• There is a need for interpretable and high-performance models for PTM prediction.

Purpose of the Study:

• To develop a novel, interpretable method for predicting protein PTMs.
• To create high-performance 'white box' classifiers for PTM prediction using pattern discovery and decision trees.
• To validate the method's efficacy on initiator methionine cleavage (IMC) and N-terminal acetylation (N-Ac).

Main Methods:

• A novel algorithm combining C4.5 decision tree algorithm with genetic algorithms for pattern discovery.
• Development of decision tree-based classifiers for PTM prediction.
• Application and testing of the method on initiator methionine cleavage (IMC) and N-terminal acetylation (N-Ac) datasets.

Main Results:

• The developed classifiers demonstrate high performance, with cross-validated MCC of 0.83 for IMC and 0.65 for N-Ac on eukaryotic proteins.
• The method outperforms state-of-the-art models in predicting substrates for N-terminal acetyltransferase B and C.
• Analysis of the IMC prediction model for Homo sapiens reveals the extraction of known biological facts, confirming the 'white box' nature of the models.

Conclusions:

• The novel pattern discovery and decision tree-based method provides high-performance, interpretable classifiers for PTM prediction.
• The 'white box' models facilitate biological knowledge extraction and validation.
• The developed predictors for IMC and N-Ac are publicly available, supporting further research.