DDAP: docking domain affinity and biosynthetic pathway prediction tool for type I polyketide synthases

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Type I modular polyketide synthases (PKS) are crucial for producing diverse bioactive compounds.
• Predicting the specific ordering of proteins and substrates in PKS pathways remains a significant challenge.
• Existing computational tools often rely on large databases or provide less intuitive outputs.

Purpose of the Study:

• To develop and evaluate DDAP, a machine learning-based tool for predicting biosynthetic pathways of type I modular PKS.
• To improve the accuracy of protein and substrate ordering within PKS pathways.
• To establish the first comprehensive database of type I modular PKS docking domain information.

Main Methods:

• Developed DDAP, a machine learning algorithm focusing on module docking domain (DD) affinity prediction.
• Utilized a hold-out testing dataset to evaluate prediction performance.
• Established a novel database for type I modular PKS, annotating docking domain information.

Main Results:

• Achieved an Area Under the ROC Curve (AUC) of 0.88 for DD affinity prediction.
• Reached a Mean Reciprocal Ranking (MRR) of 0.67 for pathway prediction.
• DDAP demonstrates high efficiency, intuitive probability-based outputs, and competitive performance against rule-based algorithms.

Conclusions:

• DDAP is the first machine learning algorithm for type I PKS DD affinity and pathway prediction.
• The tool offers advantages in efficiency and output interpretability compared to existing methods.
• The accompanying database provides valuable annotated information for bacterial PKS pathways.