You could create a local blast database using the reference fasta sequence and blast known genes of interest against that sequence to see if orthologs might be present as a simple exploration. Tutorial here. But without any gene predictions, i doubt the protein sequences you have will be useful with this method.

If you're up for it, you could run the reference sequence through a few rounds of gene predictions like Augustus, but this would be a considerable amount of work.

I'd start by simply annotating it yourself. If its a well conserved gene it'll be easy enough to pull out.

If there are good reference annotations you can use, you can usually feed annotators a list of 'reference' proteins.