I actually have an enzyme sequence from my genome annotation. I was thinking when can we say that we have a new protein based on BLAST search. How much similarity is needed to say that? below...?
It depends on what type of organism, what type of enzyme, and many other factors. A new DNA polymerase in a human genome is a bit different than a new GTPase or protein kinase in a plant genome. Even in the human genome there are thousands of potential protein coding regions which have not been definitively proven to be functional. Most complete genomes from other organisms likewise have a majority of open reading frames annotated only as "putative kinase-like protein" or something like that.
But a BLAST search with this protein sequence makes it seem clear that this is a toxic anion resistance protein. More follow-up would be needed to find out what has been studied with this gene/protein. There is no set rule for determining when a newly sequenced gene/protein should be give a new name, vs just being named as a member of a previously described group of genes/proteins.