PINC: A Tool for Non-Coding RNA Identification in Plants Based on an Automated Machine Learning Framework.

Area of Science:

• Plant molecular biology
• Bioinformatics
• Computational biology

Background:

• Non-coding RNAs (ncRNAs) are vital regulators of plant biological processes, including nutrient homeostasis, development, and stress responses.
• Accurate identification of ncRNAs is essential for functional characterization.
• Existing machine learning tools for ncRNA identification lack plant-specific applications.

Purpose of the Study:

• To develop an automated machine learning tool for accurate identification of plant non-coding RNAs (ncRNAs).
• To provide a dedicated computational resource for plant ncRNA discovery and annotation.

Main Methods:

• Extraction of 91 sequence-based features from RNA sequences.
• Feature selection using F-test and variance threshold to identify 10 key features.
• Model training with the AutoGluon framework on datasets from four plant species.
• Validation on nine independent test sets.

Main Results:

• The developed tool, PINC, achieved high accuracy in identifying plant ncRNAs, ranging from 92.74% to 96.42%.
• PINC demonstrated superior performance compared to existing tools (CPC2, CPAT, CPPred, CNIT), outperforming them in at least five evaluation metrics.
• The tool effectively identified and annotated novel ncRNAs across different plant species.

Conclusions:

• PINC is a robust and accurate tool for identifying plant non-coding RNAs.
• This tool is expected to significantly advance research in plant ncRNA biology.
• PINC facilitates the discovery and functional annotation of novel ncRNAs in plants.