RNACOREX - RNA coregulatory network explorer and classifier.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Micro-RNAs (miRNAs) and messenger RNAs (mRNAs) interactions are crucial in disease development.
• Identifying reliable miRNA-mRNA regulatory networks associated with diseases is challenging due to false positives and lack of user-friendly tools.

Purpose of the Study:

• Introduce RNACOREX, a Python package for exploring and classifying RNA coregulatory networks.
• Enable researchers to identify disease-associated miRNA-mRNA networks and classify new expression data.

Main Methods:

• RNACOREX integrates structural database information with expression data analysis.
• Utilizes conditional mutual information to infer reliable miRNA-mRNA interactions.
• Employs Conditional Linear Gaussian (CLG) classifiers for prediction and network validation.

Main Results:

• RNACOREX was tested on 13 The Cancer Genome Atlas Program databases.
• Generated post-transcriptional coregulatory networks and assessed classification performance for tumor types.
• Achieved competitive predictive performance, comparable to existing algorithms, with interpretable graph-based models.

Conclusions:

• RNACOREX provides an effective and interpretable tool for analyzing disease-associated RNA coregulatory networks.
• The package facilitates classification of new samples and validation of inferred biological networks.
• Highlights the potential of miRNA-mRNA interactions in cancer patient classification and survival prediction.