Event Analysis: Using Transcript Events To Improve Estimates of Abundance in RNA-seq Data

Area of Science:

• Computational Biology
• Genomics
• Transcriptomics

Background:

• Alternative splicing generates multiple transcripts and protein isoforms from a single gene.
• Estimating transcript abundance from short sequencing reads is challenging, leading to inaccurate transcript identification.
• Existing methods struggle with novel junction detection and accurate isoform quantification.

Purpose of the Study:

• To develop and validate the Event Analysis (EA) approach for accurate transcript and isoform abundance estimation.
• To improve the detection of novel splice junctions and annotated junctions.
• To enhance the stability and reliability of isoform abundance estimates.

Main Methods:

• Developed the Event Analysis (EA) approach to project transcripts onto the genome and catalog splice junctions and exons.
• Assembled all possible logical junctions into a comprehensive catalog.
• Filtered transcripts based on detected event proportions and coverage before quantitation.

Main Results:

• EA demonstrated superior efficiency in detecting novel junctions compared to splice-aware mapping methods.
• EA identified 99.8% of true transcripts, significantly outperforming iReckon (82%).
• EA detected 60% of novel exon combinations and ~5,000 annotated junctions missed by STAR, improving PacBio Iso-seq data analysis.
• Filtering transcripts using EA improved isoform abundance estimates, showing higher correlation between replicates and reduced variability in a Type 1 Diabetes (T1D) RNA-seq study.

Conclusions:

• The Event Analysis (EA) approach provides a more accurate and stable method for isoform abundance estimation.
• EA's focus on individual transcriptional events and simple filtering rules significantly improves transcript quantification without ancillary data.
• This method offers a robust solution for analyzing transcriptomes, particularly in studies with limited data types.