miPEPPred-FRL: A Novel Method for Predicting Plant MiRNA-Encoded Peptides Using Adaptive Feature Representation...

Area of Science:

• Molecular Biology
• Genomics
• Bioinformatics

Background:

• MicroRNAs (miRNAs) are crucial regulators in organisms.
• Small open reading frames (sORFs) in primary miRNAs (pri-miRNAs) encode miPEPs.
• Plant miPEPs enhance miRNA activity by boosting pri-miRNA transcription, affecting plant traits.

Purpose of the Study:

• To develop an efficient computational method for large-scale identification of plant miPEPs.
• To address the time-consuming and expensive nature of experimental miPEP identification.
• To introduce the first specialized computational tool for miPEP prediction.

Main Methods:

• Proposed miPEPPred-FRL, a predictor using an adaptive feature representation learning framework.
• Developed a feature transformation module with novel feature and classifier selection methods.
• Employed an improved cascade architecture for enhanced feature augmentation.

Main Results:

• The feature selection and classifier selection methods improved feature representation.
• miPEPPred-FRL demonstrated high accuracy on independent testing data.
• The predictor proved to be a reliable tool for miPEP identification.

Conclusions:

• miPEPPred-FRL offers a valuable and efficient computational approach for identifying plant miPEPs.
• The developed framework enhances feature representation for improved prediction accuracy.
• This tool facilitates large-scale miPEP discovery, aiding in the study of plant trait regulation.