Circ-LocNet: A Computational Framework for Circular RNA Sub-Cellular Localization Prediction.

Area of Science:

• Molecular Biology
• Bioinformatics
• Genomics

Background:

• Circular RNAs (circRNAs) are novel non-coding RNAs with largely unknown functions.
• Understanding circRNA sub-cellular localization is crucial for elucidating their roles in physiological processes, disease, and drug development.
• Current methods for detecting circRNA location rely on time-consuming wet experimental approaches.

Purpose of the Study:

• To develop a computational framework, Circ-LocNet, for precise prediction of circRNA sub-cellular localization.
• To evaluate the effectiveness of various sequence descriptors and machine learning classifiers for circRNA localization prediction.
• To explore the impact of sequence descriptor fusion on prediction accuracy.

Main Methods:

• Development of Circ-LocNet, a machine learning framework for circRNA sub-cellular localization.
• Extrinsic evaluation of 7 residue frequency-based, residue order and frequency-based, and physio-chemical property-based sequence descriptors.
• Utilized five widely used machine learning classifiers and assessed K-order sequence descriptor fusion (2nd to 7th order).

Main Results:

• Standalone residue frequency-based sequence descriptors with tree-based classifiers showed suitability for predicting circRNA sub-cellular localization.
• K-order heterogeneous sequence descriptors fusion combined with tree-based classifiers achieved the most accurate predictions.
• Circ-LocNet provides a robust computational baseline for novel circRNA localization determination.

Conclusions:

• Computational prediction of circRNA sub-cellular localization is feasible and accurate using machine learning.
• Tree-based classifiers and sequence descriptor fusion are key to enhancing prediction accuracy.
• Circ-LocNet facilitates future research into circRNA functions, disease associations, and therapeutic strategies.