ESPRESSO: Robust discovery and quantification of transcript isoforms from error-prone long-read RNA-seq data

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Long-read RNA sequencing (RNA-seq) offers insights into transcriptome variation and full-length transcript isoforms.
• High error rates in current long-read platforms challenge accurate transcript characterization.

Purpose of the Study:

• To develop ESPRESSO, a computational tool for robust discovery and quantification of transcript isoforms from error-prone long reads.
• To improve splice junction identification and transcript isoform detection using individual read error profiles.

Main Methods:

• ESPRESSO jointly analyzes all long reads aligned to a gene.
• It incorporates individual read error profiles to enhance accuracy.
• The tool was validated on synthetic spike-in and human RNA samples.

Main Results:

• ESPRESSO demonstrated superior performance in both transcript isoform discovery and quantification compared to contemporary tools.
• Analysis of ~1.1 billion nanopore RNA-seq reads from human samples was conducted.
• The study generated a valuable dataset for transcriptome research.

Conclusions:

• ESPRESSO provides a robust solution for analyzing error-prone long-read RNA-seq data.
• The tool enhances the study of transcript isoforms and the overall RNA repertoire.
• ESPRESSO and its associated dataset serve as a significant resource for eukaryotic transcriptome research.