LSAP: A Machine Learning Method for Leaf-Senescence-Associated Genes Prediction.

Area of Science:

• Plant Biology
• Genomics
• Computational Biology

Background:

• Plant leaves are crucial for energy conversion and food production.
• Premature leaf senescence negatively impacts crop yield and quality.
• Identifying senescence-associated genes (SAGs) is vital for understanding and managing plant aging.

Purpose of the Study:

• To develop the first computational approach for predicting plant leaf SAGs using sequence data.
• To create a user-friendly tool for SAGs prediction and annotation.
• To provide comprehensive SAGs data for a wide range of plant species.

Main Methods:

• Collected 5853 genes from a leaf senescence database.
• Developed and optimized prediction models using Support Vector Machine (SVM) and XGBoost algorithms.
• Utilized SVM-PCA-Kmer-PC-PseAAC model for best performance, leading to the SAGs_Anno tool.

Main Results:

• The SVM-PCA-Kmer-PC-PseAAC model achieved high performance (F1=0.866, Acc=0.862, ROC=0.922).
• Identified 1,398,277 SAGs across 3,165,746 gene sequences from 83 plant species.
• Observed a higher SAGs percentage in leafy species compared to leafless ones.

Conclusions:

• The SAGs_Anno tool and Leaf SAGs Annotation Platform offer valuable resources for plant senescence research.
• This study provides a novel computational method for predicting SAGs.
• The findings contribute to a deeper understanding of plant leaf senescence mechanisms and potential applications in agriculture.