PredcircRNA: computational classification of circular RNA from other long non-coding RNA using hybrid features

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Circular RNAs (circRNAs) are a significant class of long non-coding RNAs (lncRNAs) involved in gene regulation.
• circRNAs function as miRNA sponges, influencing gene expression.
• Identifying circRNAs from RNA-sequencing data is crucial for understanding their biological roles.

Purpose of the Study:

• To develop a machine learning approach for distinguishing circRNAs from other lncRNAs.
• To identify discriminative features for circRNA classification.
• To provide a computational tool for circRNA transcript identification.

Main Methods:

• Employed a machine learning approach named PredcircRNA.
• Utilized multiple kernel learning to fuse heterogeneous features.
• Extracted features including graph properties, conservation, sequence composition, repeats, SNP density, and ORFs.

Main Results:

• Achieved an accuracy of 0.778 and MCC of 0.554 in classifying circRNAs.
• Identified conservation features and GTAG motifs as discriminative.
• Demonstrated the effectiveness of the multiple kernel learning framework for feature integration.

Conclusions:

• PredcircRNA accurately distinguishes circRNAs from other lncRNAs using integrated features.
• The study highlights the importance of conservation and sequence motifs in circRNA identification.
• The PredcircRNA tool is available for researchers to identify circRNA transcripts.