DCGR: feature extractions from protein sequences based on CGR via remodeling multiple information.

Area of Science:

• Bioinformatics
• Computational Biology
• Structural Biology

Background:

• Protein feature extraction is crucial for sequence similarity analysis, structure, function, and interaction prediction.
• Graphical representation methods are effective but often limited by design.
• Existing methods for protein feature extraction have limitations.

Purpose of the Study:

• Introduce DCGR, a novel method for protein sequence feature extraction.
• Utilize chaos game representation (CGR) for enhanced feature extraction.
• Improve upon existing limitations in protein feature extraction techniques.

Main Methods:

• Construct CGR curves based on amino acid physicochemical properties.
• Convert CGR curves into multi-dimensional feature vectors using point distributions.
• Develop a novel method called DCGR for feature extraction.

Main Results:

• DCGR significantly outperforms state-of-the-art feature extraction methods.
• The method was tested on five diverse datasets.
• Demonstrated superior performance in protein sequence analysis.

Conclusions:

• DCGR is a powerful tool for extracting effective protein sequence features.
• DCGR is important for protein sequence similarity, interaction, and function prediction.
• The method is freely available for use.