A novel computational framework for genome-scale alternative transcription units prediction

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Alternative transcription units (ATUs) are crucial for bacterial gene regulation but challenging to identify comprehensively.
• Genome-scale ATU identification is vital for understanding bacterial pathogenesis and disease emergence.
• Experimental methods for ATU identification are limited by complexity and dynamic gene expression.

Purpose of the Study:

• To introduce SeqATU, the first computational framework for genome-scale ATU prediction using next-generation RNA-Seq data.
• To develop a robust method for identifying ATUs that overcomes experimental limitations.
• To enhance the understanding of bacterial transcriptional mechanisms and regulatory networks.

Main Methods:

• Development of a computational framework, SeqATU, for ATU prediction.
• Utilizing a convex quadratic programming model to optimize expression combinations of ATUs.
• Validation using two independent RNA-Sequencing datasets from Escherichia coli.

Main Results:

• SeqATU achieved high prediction accuracy (precision 0.77/0.74, recall 0.75/0.76) compared to third-generation sequencing data.
• Predicted ATUs showed a significantly higher proportion of genes with documented transcription factor binding and termination sites.
• Functional enrichment analyses (GO, KEGG) revealed stronger functional relatedness between genes within the same predicted ATUs.

Conclusions:

• SeqATU provides a reliable and efficient method for genome-scale ATU prediction.
• The findings highlight the functional coherence of genes within identified ATUs.
• SeqATU is expected to advance the study of bacterial transcription and facilitate the reconstruction of genome-scale transcriptional regulatory networks.