Historically, RefSeq bacterial genomes annotation was propagated from INSDC submissions, when available, or generated using different versions of NCBI's Prokaryotic Genome Annotation Pipeline (which is also offered as a service for GenBank submissions). This resulted in accumulated inconsistencies in both structural and functional annotation in the RefSeq prokaryotic dataset. Over the past two years NCBI improved several aspects of the Prokaryotic Genome Annotation Pipeline to increase capacity and further standardize annotation rules. Our pipeline combines a gene calling algorithm, GeneMarkS+ (49,50), with an alignment-based gene detection approach and is capable of annotating both complete and draft WGS genomes. The pipeline currently predicts protein-coding genes, structural RNAs (5S, 16S and 23S), tRNAs and small non-coding RNAs.
