Welcome to the big leaves: best practices for improving genome annotation in non-model plant genomes

Abstract

Premise of the study: Robust standards to evaluate quality and completeness are lacking for eukaryotic structural genome annotation. Genome annotation software is developed with model organisms and does not typically include benchmarking to comprehensively evaluate the quality and accuracy of the final predictions. Plant genomes are particularly challenging with their large genome sizes, abundant transposable elements (TEs), and variable ploidies. This study investigates the impact of genome quality, complexity, sequence read input, and approach on protein-coding gene prediction. Methods: The impact of repeat masking, long-read, and short-read inputs, de novo, and genome-guided protein evidence was examined in the context of the popular BRAKER and MAKER workflows for five plant genomes. Annotations were benchmarked for structural traits and sequence similarity. Results: Benchmarks that reflect gene structures, reciprocal similarity search alignments, and mono-exonic/multi-exonic gene counts provide a more complete view of annotation accuracy. Transcripts derived from RNA-read alignments alone are not sufficient for genome annotation. Gene prediction workflows that combine evidence-based and ab initio approaches are recommended, and a combination of short and long-reads can improve genome annotation. Adding protein evidence from de novo or genome-guided approaches generates more false positives as implemented in the current workflows. Post-processing with functional and structural filters is highly recommended. Discussion: While annotation of non-model plant genomes remains complex, this study provides recommendations for inputs and methodological approaches. We discuss a set of best practices to generate an optimal plant genome annotation, and present a more robust set of metrics to evaluate the resulting predictions.

Haplotype‐resolved genome assembly of Populus tremula × P. alba reveals aspen‐specific megabase satellite DNA

“…We also obtained a mono:multi‐exonic gene ratio of 0.24 and an annotation rate of 82–84% based on reciprocal BLAST against NCBI RefSeq Plant or UniProt databases for both haplotypes (Table S1). These results met the recommended metrics for high‐quality genome annotation best practices (Vuruputoor et al., 2023).…”

“…The gene number discrepancies between Populus genomes are attributable in part to lineage‐specific genes (Vuruputoor et al., 2023; Yates et al., 2021) and to copy number variation of gene family members, especially tandem gene duplicates (Żmieńko et al., 2014). Examples of tandem gene copy number variation between the two 717 haplotypes have recently been reported and include both structural and regulatory genes (Bewg et al., 2022; Chen et al., 2023).…”

GALBA: Genome Annotation with Miniprot and AUGUSTUS

“…Both GALBA and BRAKER2 tend to heavily overpredict single-exon genes, most likely a result of incorrectly splitting genes. For plants, a desired mono-to multi-exonic gene ratio of 0.2 was recently postulated by [44]. This particular ratio certainly does not hold for non-plant species, and also the reference annotations of plants used in this manuscript often deviated from that recommendation.…”

“…For Coix aquatica, we used the poales odb10. Further, we report basic metrics such as the number of predicted genes, the number of transcripts, the recently suggested mono-exonic to multi-exonic gene ratio [44], and the maximum number of exons per gene across all predicted genes.…”

Making sense of complexity: Advances in bioinformatics for plant biology

“…Despite advances in genome assembly methods, genome annotation remains one of the most challenging bottlenecks facing plant genome science, with intron length variation, divergent TE dynamics, and low sequence conservation hampering the annotation efforts of non‐model genome projects. In their contribution, Vuruptoor et al (2023) address the need to improve quantification of structural genome annotation methods, employing a mixture of existing and emerging metrics to benchmark genome annotation methods. They approach the issue in a robust manner by using a broad diversity of taxa with challenging genomic features such as variable ploidy, high TE content, and large genomes.…”