Computational Advances in ncRNA Discovery: Technical Insight Report

Author: S. M. Hesam Hosseini J.

Date: 3/30/2025

Languages Used: Python, R, Shell

Executive Summary

This report synthesizes findings from ten recent research articles in the field of non-coding RNA (ncRNA) discovery and functional annotation. These works collectively address challenges such as high false positive rates, noisy data, and limited experimental validation in large genomic datasets. Highlighted approaches span structure-guided search algorithms, modular circRNA design, RNA modification mapping, and advanced machine learning and graph-based prediction methods. Together, these tools inform a robust framework for developing next-generation RNA discovery pipelines.

Key Literature Highlights

Hansen (2018)

• Evaluated multiple circRNA prediction tools (e.g., CIRCexplorer, CIRI, MapSplice).
• Combining algorithms significantly reduced false positive rates.
• RNaseR sensitivity used to validate predictions.

Chantsalnyam et al. (2021)

• Developed ncRDense, a DenseNet-based deep model for ncRNA classification.
• Combined sequence features with RNA secondary structure encodings.

Zhu et al. (2021)

• Introduced IPCARF model (IPCA + Random Forest) to predict lncRNA-disease associations.
• Demonstrated robustness and interpretability using grid search and 10-fold cross-validation.

Zhang et al. (2022)