Usually if you have genome assembly then you have to run gene prediction first(you can use gene prediction tools such as augustus, GeneMark, Glimmer, MAKER etc.). So at the end of gene prediction, you will get gene set for your assembly then you can just perform sequence alignment using any sequence alignment tools like blast against any protein sequence database(let's say UniProt, plant refseq, plant nrdb etc). Using obtained database hits id you can find out respective annotations lets say KEGG pathways and Gene Ontology etc. using appropriate resources.
There are some paid software like blast2go for annotation and direct KEGG and GO mapping.
OR in your case, you can select the related plant genome database and do the same.
it all depends a little on your expertise and/or your ultimate goal.
if you want to have a quick look at the gene content, then tools like Maker, Maker-P etc will do. They are more of less "plug-and-play" kinda solutions without much expertise required
if on the other hand you're looking for a high quality annotation then you're likely better of with software like Augustus. EuGene, ...
The "downside" of these is that they will require a certain amount of skill to run and especially train/optimise them
I would recommend this review which was really helpful to me for the genome annotation of a fly's species.
I will give you a few tips that I learned from that process:
Remember to first mask repetitive regions, as they may be misannotated as any other genetic elements than that.
Pipelines like MAKER are quite straightforward in terms of being quite complete and not having to merge different file outputs from many software, but I did the latter to get a more personalized pipeline, and I finally merged all together using EVidenceModeler.
There are some relatively new annotation software that annotate based on an evolutionary close organism annotation, which I would recommend if such a well-studied species exist, as it would get you most of the annotation correctly; especially if your annotation would be based mostly on de novo predictions because of having a small quantity of well described OMICS data (mRNA, proteins, etc.)
Seems like I am a bit late to the party!! But you can check out our new genome annotation tools finder. It is designed to work with short-read RNA-Seq data and protein sequences. finder runs BRAKER2 to obtain predicted genes. Please check out the paper in FINDER. You can find the software in GitHub.