Discriminating cirRNAs from other lncRNAs using a hierarchical extreme learning machine (H-ELM) algorithm with...

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Circular RNAs (cirRNAs) and long non-coding RNAs (lncRNAs) are non-coding RNAs involved in crucial biological processes.
• Distinguishing cirRNAs from lncRNAs is challenging due to overlapping characteristics.
• Understanding their differences is vital for advancing RNA biology research.

Purpose of the Study:

• To identify key features differentiating cirRNAs from lncRNAs using computational approaches.
• To develop an effective classification model for distinguishing between cirRNAs and lncRNAs.
• To uncover potential indicators for cirRNA versus lncRNA identification.

Main Methods:

• Utilized a dataset of cirRNAs and lncRNAs with 188 features including graph, sequence, and conservation properties.
• Applied Minimum Redundancy Maximum Relevance (mRMR) for feature selection.
• Employed Incremental Feature Selection and a hierarchical extreme learning machine for model construction.

Main Results:

• Developed a classification model achieving 78.9% accuracy, 0.703 sensitivity, 0.850 specificity, and a 0.561 Matthews correlation coefficient.
• Identified 16 key features, with RNA sequence and structure emerging as the most significant discriminators.
• Highlighted the importance of evolutionary conservation and sequence features in distinguishing RNA types.

Conclusions:

• Computational methods can effectively differentiate cirRNAs from lncRNAs.
• RNA sequence and structure are critical features for distinguishing between cirRNAs and lncRNAs.
• The developed model provides a valuable tool for classifying non-coding RNAs.