On de novo Bridging Paired-end RNA-seq Data

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• High-throughput RNA sequencing (RNA-seq) often yields paired-end reads with unsequenced internal fragments.
• Reconstructing full-length RNA fragments from these paired ends without a reference genome (de novo bridging) is crucial for enhanced transcript analysis.
• Challenges include alternative splicing, transcript noise, and sequencing errors, making accurate de novo bridging difficult.

Approach:

• Developed a novel truncated Dijkstra's algorithm tailored for de novo bridging in compacted de Bruijn graphs (cdBGs).
• Implemented a shortest path tree reuse strategy to optimize computational efficiency, avoiding redundant calculations.
• The approach enables scalable processing of paired-end RNA-seq reads within large-scale cdBGs.

Key Points:

• Demonstrated the feasibility of computationally bridging paired-end RNA-seq reads without a reference genome.
• The developed algorithms are scalable and efficient, handling millions of vertices in cdBGs.
• Experimental results show a high degree of accuracy in reconstructing full-length RNA fragments.

Conclusions:

• The new de novo bridging method significantly enhances the utility of short-read RNA-seq data.
• This approach facilitates more accurate transcript assembly, quantification, and differential splicing analysis.
• The freely available tool (https://github.com/Shao-Group/rnabridge-denovo) supports broader adoption in RNA-seq research.