Merging short and stranded long reads improves transcript assembly.

Area of Science:

• Transcriptomics
• Genomics
• Bioinformatics

Background:

• Short-read RNA sequencing offers high depth but struggles with transcript termini and segmentation.
• Long-read RNA sequencing captures full-length isoforms but suffers from low depth and cDNA synthesis artifacts.
• Integrating both platforms is challenging due to inherent limitations and underdeveloped methods.

Purpose of the Study:

• To critically compare existing RNA sequencing assembly methods.
• To develop an integrative approach for characterizing low-abundance long noncoding RNA (lncRNA) transcripts.
• To enhance the accuracy and sensitivity of full-length transcript assembly.

Main Methods:

• Comparative analysis of short-read and long-read sequencing assembly methods.
• Development of a computational pipeline to 'strand' long-read cDNA libraries.
• Benchmarking a hybrid assembly approach integrating both sequencing types.
• Application to challenging low-abundance, under-annotated lncRNA datasets.

Main Results:

• Identified severe limitations in short-read (ambiguous ends, segmentation) and long-read (low depth, artifacts) sequencing.
• Developed a stranding pipeline to correct long-read mapping and assembly errors.
• Demonstrated that the hybrid approach significantly improves sensitivity and accuracy of full-length transcript assembly.
• Successfully resolved segmentation and depth issues for lncRNA transcript assembly with precise 5' and 3' ends.

Conclusions:

• The developed hybrid workflow effectively overcomes limitations of individual sequencing platforms.
• This approach enables superior demarcation of transcript ends and refined isoform structures.
• The method enhances differential gene expression analysis and molecular manipulation of transcripts.