Population-scale interpretation of RNA isoform diversity enabled by Isopedia

Area of Science:

• Genomics
• Transcriptomics
• Bioinformatics

Background:

• Alternative splicing creates complex transcriptomes, but reference annotation gaps inflate novel transcript identification in RNA-Seq studies.
• Existing annotation methods struggle with accuracy, leading to over 20-70% of transcripts being labeled as novel.

Purpose of the Study:

• To introduce Isopedia, a novel data structure for reference-agnostic isoform annotation.
• To establish a population-scale catalog of human transcriptome isoforms using long-read sequencing data.
• To differentiate true biological isoforms from technical noise and improve transcriptome interpretation.

Main Methods:

• Developed an expandable data structure for reference-agnostic isoform annotation.
• Generated a catalog of 1,007 long-read datasets across 37 biological contexts.
• Implemented an evidence-weighted annotation approach based on isoform frequency.

Main Results:

• Isopedia significantly reduced apparent isoform novelty by up to 26-fold in HG002 benchmarks.
• Achieved over 95% annotation rate, even for low-abundance isoforms.
• Demonstrated utility in analyzing complex genomic regions like pseudogenes and gene fusions.

Conclusions:

• Isopedia transforms isoform discovery into a systematic interpretation of the human transcriptome.
• Provides crucial frequency-based context to distinguish functional isoforms from noise.
• Establishes a foundation for advanced clinical and functional RNA research.