Utilization and Improvement Of Existing Annotation and The Incorporation Of New Gene Models In The Medicago truncatula Genome Using The MAKER Annotation Pipeline

We have used the MAKER annotation pipeline to annotate the Mt4.0 release of the Medicago truncatula genome.  There have been major improvements to the pseudomolecule build in the new release.  Compared with Mt 3.5, which contained about 250 Mb of sequence, the Mt 4.0 release contains ~360 Mb of sequence and incorporates essentially all of the previously unanchored short whole genome shotgun contigs.  While approximately 90% of the Mt3.5 gene models can be directly “lifted over” onto the Mt4.0 assemblies, nearly one third of these gene models come from short contigs.  These genes represent ab initio predictions without transcript or protein support and are often incomplete.  Therefore, a de novo annotation of the Mt4.0 release of the M. truncatula genome was required. To maintain continuity for the Medicago research community, we used MAKER to transfer and preserve the Mt3.5 Eugene-based models to the Mt4.0 release wherever possible.  In addition to considering the Mt3.5 gene models, MAKER makes use of FgenesH, GeneMark and Augustus gene predictions, and promotes the best ab initio prediction to an annotation for each locus.  Additionally, where transcript evidence supports the presence of UTRs, MAKER modifies the best gene model to include UTR sequences.  Here, we provide an overview of the Mt4.0 annotation effort and describe how these annotations were generated by MAKER.  Furthermore, we compare the Eugene-based gene models from Mt3.5 and the MAKER-based gene models from Mt4.0 as well as the MAKER annotations for the Mt4.0 pseudomolecules generated with and without the benefit of the Mt3.5 Eugene-based gene models.