We could refine the final annotation doing a final BLAST with the sequences of the final selected predicted genes. It will warrant having the most similar protein as the annotator of each predicted gene.
In this case the queries would be the predicted genes (in nucleotides because the non-canonical genes have not protein sequence) and the database would be even all bacterial proteins. BLASTX with only one hit.
This phase would be done after obtain predicted genes and before obtain the output files with the Uniprot annotations.
We could refine the final annotation doing a final BLAST with the sequences of the final selected predicted genes. It will warrant having the most similar protein as the annotator of each predicted gene.
In this case the queries would be the predicted genes (in nucleotides because the non-canonical genes have not protein sequence) and the database would be even all bacterial proteins. BLASTX with only one hit.
This phase would be done after obtain predicted genes and before obtain the output files with the Uniprot annotations.