Nucleotide-level distance metrics to quantify alternative splicing implemented in TranD

Area of Science:

• Comparative genomics
• Transcriptomics
• Bioinformatics

Background:

• Advances in transcriptome sequencing and gene prediction enable cross-species transcription comparisons.
• Existing methods lack precise quantification of structural differences in transcript models.

Purpose of the Study:

• To develop a mathematical framework and software tool for precise comparison of transcript models.
• To quantify structural variations and calculate nucleotide-level distances between transcript models.

Main Methods:

• Developed a mathematical framework to calculate transcript complexity and compare models.
• Implemented metrics for intron retention, splice site variation, and alternative UTRs.
• Created the TranD Python package (PyPi) for analyzing transcriptomes (1GTF) and comparing transcriptomes (2GTF).

Main Results:

• TranD enables quantitative comparisons of transcript models, aiding in evaluating annotation accuracy and prediction tools.
• Analysis in multiple species revealed frequent co-occurrence of alternative exons with alternative splice sites.
• Identified shared and unique transcript models between RefSeq and Ensembl, with empirical support for both.

Conclusions:

• TranD provides a precise method for comparing transcript models, supporting annotation refinement.
• Long-read RNA-seq data suggests potential under-annotation in species like D. melanogaster and D. simulans.
• Recommends combining transcript references for improved genomic annotation.