question about the busco version in funannotate

[dingxm]

Hi ,
I have plant genome which is around 1.5Gb. I do not have RNAseq data. I tried the funannotate for the gene prediction. It did a pretty good job. I have checked the busco score of the genome which is around 98% complete using Busco V5.7. When I run the funannotate, it showed the busco score is around 87% at the genome level. Is this because the version of busco in funannotate is V2 not V5? The busco score for the protein predicted by funannotate is around 82% even I used Busco V5.7. Is that because when funannotate perform the gene prediction only used 87% genes not all the busco genes for the augustus training. If we used all the busco genes for the augustus training, will the busco score for the predicted protein got improved?
Thanks