We have updated the collection of representative genome assemblies for Bacteria and Archaea. As announced in April, this set is now recalculated three times a year. We selected a total of 11,727 prokaryotic assemblies to represent their respective species among the 192,000 assemblies in RefSeq. Six hundred and thirty-five species were included in the collection for the first time, while 395 organisms from undefined species (such as Bacillus bacterium) were removed. We were able to choose a higher-quality representative than in the previous set for 18% of Bacterial and Archaeal species due to improvements in the logic of the selection that is now based on the assembly length, number of pseudo CDSs called in the PGAP annotation, number of scaffolds, whether Gene IDs are available in the Gene database for the assembly that is currently representative, and type strain status. You can see the exact criteria in order of importance on the Prokaryotic RefSeq Genomes page. Now that the new selection process is in place, we expect future updates to have fewer changes. We will replace a representative only if the assembly has changed RefSeq status or if a substantially better assembly becomes available.
You can download the reference and representative set from the Assembly resource. If you are interested in the annotation on these genomes, you can limit searches to proteins annotated on representative genomes by adding “refseq_select[filter]” to any query in the Protein database. For example, you can find all proteins annotated on representative genomes in the genus Klebsiella by using the query: “Klebsiella[organism] AND refseq_select[filter]“. A BLAST database of proteins annotated on representative genomes will be coming soon. Stay tuned!
Release 3.0 of the NCBI protein family models used by the Prokaryotic Genome Annotation Pipeline (PGAP) is now available from our FTP site. You can search this collection of hidden Markov models (HMMs) against your favorite prokaryotic proteins to identify their function using the HMMER sequence analysis package.
The 3.0 release contains 17,350 models: 12,864 HMMs built at NCBI (111 more than in release 2.0) and 4,486 TIGRFAM HMMs. In addition, since release 2.0, we have assigned product names to over 2,000 Pfam HMMs, bringing the total to 6,698 Pfam HMMs with names that can be transferred by PGAP to the annotated proteins they hit. You can access a table of these product names from the release directory.Figure 1. The evidence for name assignment for type III secretion system (T3SS) translocon subunit SctB (NF038055) showing the protein matches. Species-specific names for this highly variable component of T3SS include YopD, EspB, IpaC, SipC, etc. Instead, we used the standard moniker for core genes of T3SS, Sct, Secretion and cellular translocation (PMID 26520801, PMID 9618447) providing a unified nomenclature for this secretion system. Continue reading “Updated protein family models used by PGAP available for download”→
A new release of the NCBI protein families profiles used by PGAP (the Prokaryotic Genome Annotation Pipeline) is now available. You can search this collection of Hidden Markov models (HMM) against your favorite prokaryotic proteins to identify their function using hmmer.
The HMMs are used as hints for the structural annotation of protein-coding genes in bacterial genomes and are also one of the sources for the names assigned to PGAP-annotated proteins presented in the Evidence-For-Name-Assignment comment block of RefSeq protein records (See for example, WP_004152100.1).
The collection comprises 12,753 HMMs that were built at NCBI, and 4,486 TIGRFAM HMMs whose ownership was transferred to NCBI in April 2018. In addition to the HMM profiles and seed alignments, a tab-delimited file containing the product names and other attributes added to the HMMs by curators is available.
85% of models were assigned a product name that can be transferred to proteins hit by the model.
7702 models have gene symbols.
14508 are supported by a least one publication.
6266 are assigned an Enzyme Commission number.
617 represent anti-microbial resistance proteins.
Product names added to 4,686 PFAM HMMs owned by EBI-EMBL and used for functional annotation by PGAP are also included.
A total of 57 million RefSeq prokaryotic proteins have been named based on these curated HMMs, and can be identified with the Entrez query “meta Evidence-For-Name-Assignment”[Properties] AND “Evidence Category=HMM”[Text Word]. See an example and more information on web displays of HMMs in a previous post.
A new version of the Prokaryotic Genome Annotation Pipeline (PGAP) with several important features is now available on Github.
In response to several requests we have added the option of running PGAP with Singularity, Podman or any other Docker-compatible executable you wish to use.
We have also lifted the requirement for internet access in case you have privacy concerns. To run the pipeline without internet access, set the flag
Are you unsure about the identity of organism you sequenced? We’ve added the Taxonomy-Check module to help you. This module will confirm the organism name or suggest a new taxonomic assignment through average nucleotide identity comparison with type material assemblies from GenBank. The check is currently an optional validation step prior to PGAP.
Try these new features and let us know what you think! Or submit your PGAP-annotated assembly to GenBank. And remember that if you are still improving the assembly and your genome doesn’t pass the pre-annotation validation, you can use the --ignore-all-errors flag to get a preliminary annotation.
We are making changes to the set of bacterial and archaeal RefSeq Reference and Representative assemblies in February 2020.
We will reduce the number of Reference assemblies to 15 that have annotation provided by outside experts (Table 1) and re-annotate the 105 other current Reference assemblies using the latest Prokaryotic Genome Annotation Pipeline (PGAP) software. The re-annotated assemblies will lose reference status.
We will reassess and revise the set of Representative assemblies so that there is one assembly per species to better reflect the taxonomic diversity of the RefSeq bacterial and archaeal assemblies.
Check out the latest videos on YouTube to learn how to best use NCBI graphical viewers, SRA, PGAP, and other resources.
Genome Data Viewer: Analyzing Remote BAM Alignment Files and Other Tips
This video shows you how to upload remote BAM files, and succinctly demonstrates handy viewer settings, such as Pileup display options, and highlights the very helpful tooltips in the Genome Data Viewer (GDV). There’s also a brief blog post on the same topic.
A new version of the Prokaryotic Genome Annotation Pipeline (PGAP) is now available on GitHub. This release uses a new and improved version of tRNAscan (tRNAscan-SE:2.0.4) and includes our most up-to-date Hidden Markov Model and BlastRule collections for naming proteins.
Remember that you can submit the results of PGAP to GenBank. Or, if you are still improving the assembly and your genome doesn’t pass the pre-annotation validation, you can use the –ignore-all-errors mode to get a preliminary annotation.
How does it work? Download PGAP from GitHub, provide some basic information and the FASTA sequences for your genome sequence, and run the pipeline on your own machine, compute farm or the cloud. PGAP will produce annotation consistent with NCBI’s internal PGAP. Submit the resulting annotated genome to GenBank through the genome submission portal, and get an accession back.
As with any other submitted assembly, PGAP-annotated genomes will be screened for foreign contaminants and vector sequences at submission. Any annotated assemblies that don’t pass may need to be modified. We are developing an automated process to handle these edits!
We are also working on other improvements to stand-alone PGAP such as a module for calculating Average Nucleotide Identity (ANI) to confirm the assembly’s taxonomic assignment. Stay tuned for new developments!
You can now download PGAP from GitHub and run it on your machine, compute farm or the cloud, on any public or privately-owned genome. PGAP predicts genes on bacterial and archaeal genomes using the same inputs and applications used inside NCBI. This is a great opportunity for you to try it now and send us comments (please use GitHub issues).
NCBI has been asked to take over the ownership and maintenance of the TIGRFAM collection of Hidden Markov Models (HMMs), which is widely used for the annotation of prokaryotic genomes. The TIGRFAMs are a collection of curated protein families started in 1998 at The Institute of Genomic Research (TIGR), precursor to the J. Craig Venter Institute (JCVI). This collection is publicly available under a Creative Commons license and downloadable from NCBI. We have already made hundreds of improvements to TIGRFAM names and descriptions and we will continue to make regular updates.\