I am not entirely sure what data you have and what you are looking for. I infer that you have a genome annotated through RAST and want to compare a particular ORF/gene with other taxa for which you do not presently have genomes/the gene of interest. If this is the case, I would extract your gene from RAST and BLAST it against the non-redundant database on NCBI. You can then align and infer phylogenetic reconstructions from this data or any number of other analyses.

If you have other annotated genomes you particularly wish to compare (and especially if there is other genomic analyses you will wish to perform) clustering the genes into homologous groups is an excellent first step. For this I 2nd Danielle Cloutier and recommend GET_HOMOLOGUES as an excellent and user-friendly program. You can then extract the HG(s) containing the gene(s) of interest and proceed with those sets.