PGAB: From genome sequence to draft model

PGAB stands for Prokaryotic Annotation Pipeline (PGAP)-based workflow.

Caution

This workflow is under development on the pgab-dev branch and is planned for a future update. The overview below describes the planned workflow.

Starting from a genome sequence, PGAB aims to annotate the genome and extend the protein annotations with database identifiers before generating a draft model. After each of the major stages (highlighted in green in Fig. 11) the workflow can be stopped and only the according output will be returned. Fig. 11 shows the following stages:

  1. Taxonomy check and/or genome annotation: Uses NCBI’s Prokaryotic Genome Annotation Pipeline (PGAP) with the genome FASTA file as input. This stage produces an annotated FASTA and GFF file and/or taxonomy information in XML/ TXT format depending on the run settings (tax-check = 'only': Taxonomy information in XML/TXT format, tax-check = 'continue': All files, tax-check = 'none': Annotated FASTA and GFF file).

  2. Depending on the result from stage 1 and the user-input either route A - RefSeq Lookup or route B - Annotation extension is selected. If the taxonomy check was performed and PGAP found a taxon with 100% identity and the user did not specify that DIAMOND should be run route A is chosen. In all other cases the next stage is a DIAMOND search to extend the annotation content.

    1. RefSeq Lookup: Based on the Genbank identifier with 100% ANI the corresponding RefSeq identifier is searched.

    2. Annotation extension: DIAMOND is used to obtain additional protein annotations. The DIAMOND route includes taxonomy-specific and unrestricted searches against NCBI’s nr database, SwissProt, or a user-provided database. The taxonomy-specific search is only performed if a taxid was found with PGAP or if the user provided a taxid.

  3. Draft generation: Uses either the data from NCBI based on the looked up RefSeq identifier (result from 2.A) or the annotated files from PGAP along with the extended annotations if route 2.B was used with either the CMPB or HQTB workflow to generate the draft model.

  4. Output: The output varies depending on the following cases:

    1. Only PGAP was requested: The output of PGAP is returned depending on the chosen tax-check setting.

      Important

      Special case: tax-check = 'only' returns only the taxonomy report and never performs any of the other steps.

    2. Route 2.A was used but no draft model generation requested: Only a warning is logged and the results from the steps before are returned.

    3. Route 2.B was used but no draft model generation requested: Only a mapping table in CSV format linking the annotated proteins to database identifiers is returned along with the files from the previous steps.

    4. The user specified that a draft model is generated: The draft model is returned in SBML format and a mapping table in CSV format as well as the files from the previous steps.

Planned PGAB workflow with PGAP annotation, RefSeq lookup and DIAMOND searches, CMPB or HQTB draft generation, and CSV and SBML outputs

Fig. 11 Planned PGAB workflow from a genome sequence to a draft model and a mapping table.