COYOTE: Sequence-derived structural descriptors-based computational identification of glycoproteins

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology
• Bioinformatics

Background:

• Glycoproteins are vital in numerous biological functions, including cell signaling, infection, and metastasis.
• Experimental identification of glycoproteins is resource-intensive, necessitating efficient computational approaches.
• Accurate modeling of glycosylated protein recognition aids in designing carbohydrate-derived therapeutics.

Purpose of the Study:

• To develop a novel machine learning model for accurate glycoprotein identification.
• To address limitations of existing methods by incorporating sequence-derived structural descriptors (SDSD).
• To provide a computational tool that complements experimental approaches and reduces costs.

Main Methods:

• Proposed a machine learning-based predictive model utilizing sequence-derived structural descriptors (SDSD).
• Evaluated model performance using machine learning-centric and biologically relevant metrics.
• Conducted data mining to identify key descriptors influencing glycoprotein determination.

Main Results:

• The proposed model demonstrates state-of-the-art generalization performance in glycoprotein identification.
• SDSD effectively compensates for the lack of protein 3D structures and enhances sequence-based analysis.
• Identified specific descriptors crucial for accurate glycoprotein prediction.

Conclusions:

• The novel machine learning model offers a highly accurate and efficient method for glycoprotein identification.
• The COYOTE webserver and code facilitate broader application in research and drug discovery.
• This approach reduces the reliance on costly and time-consuming experimental methods.