PAMPred: A hierarchical evolutionary ensemble framework for identifying plant antimicrobial peptides

Area of Science:

• Plant Science
• Biochemistry
• Computational Biology

Background:

• Antimicrobial peptides (AMPs) are vital for plant immunity, growth, and development.
• Wet-lab identification of plant AMPs is time-consuming and expensive.
• Accurate computational methods are needed for novel plant AMP discovery.

Purpose of the Study:

• To develop an accurate computational framework for identifying plant AMPs.
• To address challenges like class imbalance in AMP prediction.
• To provide an efficient tool for plant AMP research.

Main Methods:

• Developed PAMPred, a hierarchical evolutionary ensemble framework.
• Employed cluster-based resampling to handle class imbalance.
• Integrated sequence and physicochemical features into diverse base learners.
• Optimized ensemble weights using improved particle swarm optimization (PSO) and dynamic pruning.

Main Results:

• PAMPred demonstrated excellent prediction performance and generalization ability.
• The method significantly outperformed existing state-of-the-art techniques.
• Extensive cross-validation and independent testing validated the model's efficacy.

Conclusions:

• PAMPred is an effective auxiliary tool for identifying plant AMPs.
• The framework aids in exploring plant immune regulatory mechanisms.
• This computational approach accelerates the discovery of novel plant AMPs.