A network-based computational framework to predict and differentiate functions for gene isoforms using exon-level...

Area of Science:

• Genomics
• Bioinformatics
• Systems Biology

Background:

• Alternative splicing generates transcript and protein diversity, with isoforms performing specific biological functions.
• High-throughput sequencing enables high-resolution estimation of isoform-specific exon expression, crucial for network analysis of splice variants.

Purpose of the Study:

• To propose a novel network-based analysis framework, Iso-Net, for predicting isoform-specific functions from exon-level RNA-Seq data.
• To integrate MINet and RVNet for robust co-expression network inference across diverse data scenarios.

Main Methods:

• Developed Iso-Net, a unified framework integrating MINet and RVNet for co-expression network inference.
• Employed quantitative measures like Jaccard correlation coefficient for network analysis.
• Combined differential co-expression network analysis and Gene Ontology (GO) functional enrichment analysis.

Main Results:

• Iso-Net demonstrated superior prediction accuracy compared to existing methods, particularly with small sample sizes or disparate exon numbers between isoforms.
• Successfully predicted isoform functions and identified distinct functions within the same gene.
• Applied Iso-Net to analyze exon-level RNA-Seq data of transcription factors in human myeloid differentiation.

Conclusions:

• Iso-Net provides a powerful tool for predicting isoform-specific functions from exon-level RNA-Seq data.
• The framework enhances understanding of functional diversity arising from alternative splicing.
• Iso-Net is openly available for broader research applications.