Effectidor II: a pan-genomic AI-based algorithm for the prediction of type III secretion system effectors

Area of Science:

• Bacterial genomics
• Molecular microbiology
• Bioinformatics

Background:

• Type III secretion systems (T3SS) deliver effector proteins into host cells, influencing bacterial pathogenesis and host immune responses.
• T3SS effector (T3E) repertoires are diverse and evolve rapidly, necessitating comprehensive identification for understanding bacterial virulence and host-pathogen interactions.
• Accurate identification of T3Es is crucial for disease management and uncovering novel virulence factors.

Purpose of the Study:

• To develop an advanced web server, Effectidor II, for identifying type III effector genes.
• To enable simultaneous analysis of multiple bacterial genome sequences (pan-genomes).
• To leverage machine learning for improved T3E prediction accuracy.

Main Methods:

• Effectidor II utilizes machine learning algorithms trained on features extracted from entire pan-genome sequences.
• The web server facilitates the identification of T3Es across multiple related bacterial genomes.
• A novel T3E was discovered using the Effectidor II prediction capabilities.

Main Results:

• Effectidor II successfully predicts type III effector genes within bacterial pan-genomes.
• The study demonstrates the advantage of machine learning approaches utilizing pan-genome-wide sequence features.
• A previously unknown T3E was identified in *Xanthomonas euroxanthea*.

Conclusions:

• Effectidor II enhances the capability to identify T3Es from multiple genomes, advancing the study of bacterial virulence.
• The tool aids in distinguishing core vs. specialized T3Es, crucial for understanding pathogen evolution and disease dynamics.
• Effectidor II facilitates the discovery of novel T3Es, contributing to pathogen surveillance and control strategies.