Evolutionary and methodological considerations when interpreting gene presence-absence variation in pangenomes

Area of Science:

• Genomics
• Bioinformatics
• Plant Science

Background:

• Graph-based pangenomes are standard for genome comparisons.
• Integrating gene annotations across pangenomes into a single pangene set is challenging.
• Methodological inconsistencies and biological presence-absence variation (PAV) contribute to these challenges.

Purpose of the Study:

• To review and experimentally evaluate the sources of inconsistency in genome annotation and pangene set construction.
• To investigate these issues using cotton and soybean polyploid plant pangenomes.
• To provide recommendations for accurate pangene set integration.

Main Methods:

• Reviewed existing literature on pangenome annotation.
• Experimentally evaluated annotation inconsistency using cotton and soybean pangenomes.
• Re-annotated genomes using a single, identical pipeline.

Main Results:

• Building pangene sets across different genome resources is highly error-prone.
• PAV calculated from public annotations reflects annotation methods, not biological differences.
• Re-annotation with a unified pipeline resolved major issues, but identical sequences with different annotations persisted.
• Significant challenges remain in integrating gene model annotations.

Conclusions:

• Pangenome gene model annotations must be carefully integrated before biological inferences.
• Annotation methodology significantly impacts pangene set construction and PAV analysis.
• Standardized annotation pipelines and careful integration are crucial for reliable pangenome comparisons.