New feature extraction from phylogenetic profiles improved the performance of pathogen-host interactions

Area of Science:

• Computational Biology
• Bioinformatics
• Genomics

Background:

• Understanding pathogen-host interactions (PHIs) is crucial for deciphering molecular mechanisms between organisms.
• Experimental methods for PHI exploration are resource-intensive, necessitating efficient computational approaches.
• Existing machine learning (ML) methods for PHI prediction often rely on sequence-derived structure-based features.

Purpose of the Study:

• To develop a novel computational method for predicting PHIs.
• To improve the accuracy and efficiency of PHI prediction compared to existing methods.
• To leverage evolutionary information for enhanced PHI discovery.

Main Methods:

• Proposed a new method for extracting features from phylogenetic profiles to capture evolutionary information.
• Developed five distinct feature extraction models.
• Integrated phylogenetic profile features with structure-based information for prediction.

Main Results:

• The proposed method demonstrates superior performance compared to traditional structure-based and ML-based PHI prediction techniques.
• Combining phylogenetic profile features with structure-based information significantly enhances prediction accuracy.
• The approach facilitates the exploration of PHIs and the identification of novel biological relationships.

Conclusions:

• Phylogenetic profile features offer valuable evolutionary insights for PHI prediction.
• The developed method provides a more effective computational tool for discovering unknown biological relationships.
• The integration of diverse feature types holds promise for advancing PHI research.