# NCBI Prokaryotic Genome Annotation Pipeline


## Key Takeaways

- The NCBI Prokaryotic Genome Annotation Pipeline (PGAP) is a standardized, automated system for annotating bacterial and archaeal genomes, crucial for submission to GenBank and RefSeq. It identifies protein-coding genes using alignment-based predictors like ProSplign and Gnomon, refined by homology searches against curated protein databases such as RefSeq and Pfam.
- High-quality genome assemblies are paramount for accurate PGAP annotation; NCBI mandates a minimum of 50x coverage for short-read assemblies and recommends N50 values above 20 kb for draft genomes, with rigorous checks for contamination and completeness using tools like CheckM or BUSCO.
- PGAP performs functional annotation by assigning gene symbols, EC numbers, and Gene Ontology terms based on evidence from homology searches and Hidden Markov Models (HMMs), but novel genes without known homologs are often labeled as "hypothetical protein."
- Submission to PGAP is typically initiated via the NCBI Genome Submission Portal, requiring the creation of BioProjects and BioSamples with comprehensive metadata, or alternatively, through a command-line interface for more control.
- Post-annotation validation is essential, including comparing annotated gene counts to related genomes, verifying the presence of essential single-copy genes (e.g., RecA), and cross-referencing a subset of annotations with BLAST to confirm predicted protein functions.
- Limitations of PGAP include its dependence on reference databases, potential disruption of gene predictions in fragmented assemblies, and the inherent absence of experimental confirmation, necessitating follow-up laboratory work for definitive functional claims.

---

The NCBI Prokaryotic Genome Annotation Pipeline (PGAP) is a standardized, automated system that identifies genes, coding sequences, noncoding RNAs, and other genomic features in bacterial and archaeal genomes. This guide is for researchers, bioinformaticians, and clinical microbiologists who want to submit complete or draft prokaryotic genomes to NCBI and receive high quality, reproducible annotations. PGAP is the official annotation pipeline used by NCBI for RefSeq genomes, and understanding its workflow, decision points, and limitations will help you produce submission ready data that meets public database standards. For a foundational understanding of genome structure, see our guide on [Protein Synthesis](/blog/guides/protein-synthesis). The pipeline relies on curated protein databases, Hidden Markov Models (HMMs), and comparative genomics to produce consistent annotations across all prokaryotic submissions [NCBI Bookshelf](https://www.ncbi.nlm.nih.gov/books/).

## At a Glance

| Aspect | Description |
|--------|-------------|
| **Purpose** | Automated annotation of prokaryotic genome assemblies for public submission to GenBank and RefSeq |
| **Input** | Complete or draft genome assembly in FASTA format (contigs or scaffolds) |
| **Output** | GenBank flat file (.gbff) with annotated features, protein sequences, and evidence tags |
| **Key Databases** | RefSeq protein collection, Pfam, TIGRFAMs, and curated HMM models |
| **Submission Path** | Via NCBI Genome Submission Portal or command line with Aspera |
| **Timeframe** | 24,72 hours for most bacterial genomes |
| **Cost** | Free for academic and public use |
| **Supported Taxa** | Bacteria and Archaea (including plasmids) |

## Core Concepts and Decision Criteria

PGAP uses a tiered approach. First, it detects protein coding genes using ProSplign and Gnomon, two alignment based gene predictors, then refines predictions with homology searches against the RefSeq protein database. Noncoding RNAs are identified with Infernal using Rfam models. The pipeline also performs functional annotation by assigning gene symbols, EC numbers, and Gene Ontology terms where evidence supports them [EMBL EBI Training](https://www.ebi.ac.uk/training/). For a deeper look at how gene products are structured, review [Protein Structure](/blog/guides/protein-structure).

### Decision Points Before Submission

1. **Assembly Quality**  
   PGAP works best with high quality assemblies. NCBI requires a minimum of 50x coverage for short read assemblies and recommends N50 values above 20 kb for draft genomes. Check your assembly for contamination and completeness using tools such as CheckM or BUSCO. The Galaxy Training Network provides accessible workflows for assembly quality assessment [Galaxy Training Network](https://training.galaxyproject.org/).

2. **Genome Representation**  
   You must decide whether your genome represents a new species, a new strain, or a resequencing of an existing RefSeq entry. If your genome is from a novel species, you will need to provide taxonomic verification (16S rRNA similarity or average nucleotide identity with type strains).

3. **Submission Format**  
   PGAP is triggered automatically when you submit an assembly to GenBank via the Genome Submission Portal. Alternatively, you can run the pipeline locally using the PGAP command line tool if you need more control or frequent iterative annotations. The NCBI Bookshelf details both submission paths [NCBI Bookshelf](https://www.ncbi.nlm.nih.gov/books/).

## Practical Workflow for PGAP

Follow these steps to obtain a fully annotated prokaryotic genome from NCBI.

### Step 1: Prepare Your Assembly

Assemble your raw reads into contigs or scaffolds. Common assemblers include SPAdes, Unicycler, or SKESA. Remove adapters and quality trim reads using tools like Trimmomatic. The NCBI Sequence Read Archive (SRA) is the repository for the raw data [NCBI Sequence Read Archive](https://www.ncbi.nlm.nih.gov/sra). A typical good practice is to deposit your reads in SRA first, then link them to your assembly submission.

### Step 2: Perform Quality Checks

Run quality metrics on your assembly:
- Number of contigs (preferably < 200 for draft genomes)
- N50 > 20 kb
- Genome completeness (> 95% using CheckM)
- Check for contamination using Kraken or Kmer based tools
- Confirm correct phylogenetic placement

Several published studies demonstrate the importance of these quality steps. For example, the whole genome sequence of a superbug Escherichia coli strain KAB AI 497 described a careful assembly process that met NCBI quality thresholds before annotation [Whole genome sequence of a superbug, PubMed](https://pubmed.ncbi.nlm.nih.gov/42363042/). Similarly, the genomic characterisation of a Neisseria gonorrhoeae strain used thorough contamination checks to ensure annotation reliability [Genomic characterisation, PubMed](https://pubmed.ncbi.nlm.nih.gov/42342119/).

### Step 3: Submit to NCBI Genome Submission Portal

Create a BioProject and BioSample at NCBI. The BioProject describes your study, and the BioSample contains metadata about the organism (isolation source, geographical location, host). Then submit your assembly FASTA file. NCBI will automatically run PGAP. You can track the annotation status in the submission portal.

### Step 4: Interpret the PGAP Output

Once complete, you will download a GenBank flat file containing:
- Gene coordinates and product names
- Protein translations
- Noncoding RNA features
- Functional annotations with evidence codes

Cross check a subset of annotations manually using BLAST to confirm that predicted proteins match known sequences. Bioconductor provides R packages for parsing and analyzing GenBank files [Bioconductor](https://bioconductor.org/). For a broader biological context, the cell membrane is a critical structure whose function depends on properly annotated genes [Cell Membrane Function Biology](/blog/guides/cell-membrane-function-biology).

### Step 5: Submit to RefSeq (Optional)

If your genome passes NCBI’s internal review, it may be promoted to RefSeq status. This adds an additional layer of validation and curation. RefSeq genomes receive an NCBI accession prefix (e.g., NC_xxxxxx) and are included in the reference genome database.

## Quality Checks and Validation

After receiving your PGAP annotation, perform these validation steps:

- Check the total number of annotated genes against closely related genomes. Bacterial genomes typically contain 1,500 to 6,000 protein coding genes.
- Verify that essential single copy genes are present (e.g., RecA, ribosomal proteins).
- Look for pseudogenes or frameshifts that may indicate assembly errors.
- Compare your annotation with that of a close relative using tools like Mauve or ACT.

A published pipeline called AquaaG specifically focuses on quality assessment and annotation of genomes, providing a comprehensive framework that complements PGAP [AquaaG, PubMed](https://pubmed.ncbi.nlm.nih.gov/42199883/). Using multiple validation approaches increases confidence in your annotation.

## Common Mistakes and How to Avoid Them

1. **Submitting Low Quality Assemblies**  
   PGAP will annotate anything you submit, but poor assemblies yield fragmented gene predictions and missing features. Always assess N50 and contig counts before submission.

2. **Incorrect Taxonomic Assignment**  
   PGAP uses the submitted taxonomy to select reference databases. Misidentified taxa can lead to incorrect functional annotations. Use a reliable method like ANI or 16S rRNA comparison to confirm taxonomy.

3. **Ignoring Contamination**  
   Contaminant contigs from other organisms will be annotated as part of your genome, inflating gene counts and causing errors. Use Kraken or similar tools to screen for contamination.

4. **Neglecting Metadata Completeness**  
   Missing BioSample attributes can delay submission or result in incomplete annotation records. Provide geographic location, host, isolation source, and collection date.

5. **Overinterpreting Functional Annotations**  
   PGAP provides evidence based predictions, but many gene products are labeled “hypothetical protein” because homology is insufficient. Avoid claiming functions without experimental validation. For example, studies on Stenotrophomonas rhizophila highlighted that many predicted aflatoxin reduction genes remained hypothetical until confirmed by biochemical assays [Genomic Features of Stenotrophomonas, PubMed](https://pubmed.ncbi.nlm.nih.gov/42271034/).

## Limits and Uncertainty

PGAP is powerful but not perfect. Several limitations constrain its output:

- **Dependence on Reference Databases**  
  The pipeline relies on curated protein databases. Novel genes without known homologs will be missed or labeled as hypothetical. This is especially problematic for poorly studied lineages.

- **Incomplete Genome Assemblies**  
  Draft genomes with many contigs may disrupt gene predictions, especially for operons or genes spanning contig boundaries.

- **Misannotation of Pseudogenes**  
  PGAP may annotate fragmented genes as functional if they show partial homology, leading to over prediction of intact coding sequences.

- **No Experimental Confirmation**  
  PGAP is purely computational. Annotations are predictions, not experimental data. Functional claims require follow up laboratory work.

- **Updates and Version Changes**  
  PGAP is periodically updated. Re annotating the same assembly with a newer version may produce different results. Always record the PGAP version used in your publications.

A study on Carbapenem resistant Klebsiella pneumoniae in Iraq used PGAP annotations to identify resistance genes, but the authors noted that only a subset were confirmed by phenotypic testing [Whole genome sequencing of Klebsiella, PubMed](https://pubmed.ncbi.nlm.nih.gov/41970180/). This illustrates the importance of combining computational annotation with experimental validation.

## Frequently Asked Questions

**1. Can I run PGAP locally without an internet connection?**  
Yes. NCBI provides a command line version of PGAP that you can install via Docker. However, you need to download reference databases first, which may require several hundred gigabytes of storage. Internet is still required for initial setup and database updates.

**2. How long does PGAP take for a typical bacterial genome?**  
For a 5 Mb genome with 50 contigs, PGAP usually completes within 24 to 48 hours. Larger or more complex genomes (e.g., with many plasmids) may take up to 72 hours. You can check the status in the submission portal.

**3. What should I do if my annotation contains many hypothetical proteins?**  
This is normal for genomes from poorly studied taxa. Consider using additional annotation tools such as RASTtk or Prokka for comparison. The presence of many hypothetical proteins does not indicate a problem with your assembly or submission.

**4. Does PGAP annotate plasmids separately?**  
Yes. If you include plasmid sequences in the same assembly FASTA file as separate entries, PGAP annotates them individually. Each plasmid receives its own GenBank accession. Ensure plasmid entries have appropriate identifiers in the FASTA header.

## Related Clinical & Scientific Guides

* [Observational vs. Experimental Studies: How to Tell Them Apart](/blog/guides/observational-vs-experimental-studies-how-to-tell-them-apart)
* [Astrocyte Single Cell Rna Seq](/blog/guides/astrocyte-single-cell-rna-seq)
* [Structural Genes](/blog/guides/structural-genes)


## References and Further Reading

- NCBI Prokaryotic Genome Annotation Pipeline: Detailed technical documentation [NCBI Bookshelf](https://www.ncbi.nlm.nih.gov/books/)
- EMBL EBI Training on genome annotation: Practical modules and case studies [EMBL EBI Training](https://www.ebi.ac.uk/training/)
- Galaxy Training Network: Workflows for assembly quality assessment and annotation [Galaxy Training Network](https://training.galaxyproject.org/)
- Bioconductor: R packages for GenBank file parsing and analysis [Bioconductor](https://bioconductor.org/)
- NCBI Sequence Read Archive: Repository for raw sequencing reads [NCBI Sequence Read Archive](https://www.ncbi.nlm.nih.gov/sra)
- AquaaG pipeline: Quality assessment and annotation of genomes [AquaaG, PubMed](https://pubmed.ncbi.nlm.nih.gov/42199883/)
- Genomic features of Stenotrophomonas rhizophila: Annotation limits in functional prediction [Genomic Features, PubMed](https://pubmed.ncbi.nlm.nih.gov/42271034/)
- Whole genome sequencing of Klebsiella pneumoniae: Combining annotation with phenotypic testing [Klebsiella genome, PubMed](https://pubmed.ncbi.nlm.nih.gov/41970180/)
- Molecular characterization of Staphylococcus pseudintermedius: PGAP for clinical isolates [Staphylococcus genome, PubMed](https://pubmed.ncbi.nlm.nih.gov/42188894/)
- Genomic characterisation of Neisseria gonorrhoeae: Annotation of antimicrobial resistance genes [Neisseria genome, PubMed](https://pubmed.ncbi.nlm.nih.gov/42342119/)

## Related Articles

- [Protein Synthesis](/blog/guides/protein-synthesis)
- [Incomplete Dominance Gene](/blog/guides/incomplete-dominance-gene)
- [Cell Membrane Function Biology](/blog/guides/cell-membrane-function-biology)
- [Dna Structure](/blog/guides/dna-structure)
- [Protein Structure](/blog/guides/protein-structure)