Using multiple reference genomes to identify and resolve annotation inconsistencies.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Advances in sequencing technologies yield multiple reference genomes with differing gene model annotations.
• Split-gene misannotations, where single genes are split or multiple genes merged, impact functional predictions and expression estimates.

Purpose of the Study:

• To develop a high-throughput method for detecting and quantifying split-gene misannotations.
• To provide a robust framework for correcting these misannotations using expression data.

Main Methods:

• Pairwise comparison of genome annotations to identify potential split-gene errors.
• Utilized RNA-seq data from 10 tissues and a novel simulation framework to validate gene models.
• Developed a metric to quantify support for merging or splitting gene models.

Main Results:

• Identified several hundred potential split-gene misannotations across pairwise comparisons of maize genomes.
• These misannotations represent 3-5% of gene models in the analyzed annotations.
• The method requires minimal human interaction and is applicable to future genome assemblies.

Conclusions:

• Split-gene misannotations are frequent in maize genome annotations.
• The developed method efficiently identifies and corrects these errors.
• Accurate gene models are essential for reliable biological inference, especially in expression-based studies.