This database is derived from the Pathogen Detection Reference Gene Catalog, Pathogen Detection Reference Gene Hierarchy, and Reference HMM Catalog and is used by the Pathogen Detection isolate analysis system to provide results to the Isolates browser and MicroBIGG-E as well as the command-line version of AMRFinderPlus. The 'core' subset version focuses on acquired or intrinsic AMR gene products including point mutations in a limited set of taxa.

The 'plus' subset include a less-selective set of genes of interest including genes involved in virulence, biocide, heat, metal, and acid resistance.

The most recent database release can be found in https://ftp.ncbi.nlm.nih.gov/pathogen/Antimicrobial_resistance/AMRFinderPlus/database/latest and a log of changes with each release is available in the changes.txt file. Note that this database is compiled as part of the National Database of Antibiotic Resistant Organisms (NDARO) and more user-friendly access to the data is available at https://www.ncbi.nlm.nih.gov/pathogens/antimicrobial-resistance/. The rest of this document describes the format and structure of the database as it used by AMRFinderPlus.

Users of AMRFinderPlus or its supporting data files are cautioned that presence of a gene encoding an antimicrobial resistance (AMR) protein does not necessarily indicate that the isolate carrying the gene is resistant to the corresponding antibiotic. AMR genes must be expressed to confer resistance. An enzyme that acts on a class of antibiotic, such as the cephalosporins, may confer resistance to some but not others. Many AMR proteins reduce antibiotic susceptibility somewhat, but not sufficiently to change the isolate from "sensitive" to "intermediate" or "resistant." Meanwhile, an isolate may gain resistance to an antibiotic by mutational processes, such as the loss of porin required to allow the antibiotic into the cell. For some families of AMR proteins, especially those borne on plasmids, correlations of genotype to phenotype are much more easily deciphered, but users are cautioned against over-interpretation.

The AMRFinderPlus database is publicly available at https://ftp.ncbi.nlm.nih.gov/pathogen/Antimicrobial_resistance/ Files on the FTP site are in the structure:

|- /AMRFinderPlus
     |- data  # depricated
          |- latest
          |- YYYY-MM-DD.#
               |- fam.tab
               |- AMRProt
               |- AMR.LIB
               |- changes.txt
               |- ReferenceGeneCatalog.txt
               ...
          |- YYYY-MM-DD.#
            ...
     |- database 
         |- <database format version> e.g., 3.10
               |- YYYY-MM-DD.#
                    |- fam.tab
                    |- AMRProt
                    |- AMRProt-mutation.tab
                    |- AMRProt-suppress
                    |- AMRProt-susceptible.tab
                    |- AMR.LIB
                    |- AMR_CDS
                    |- AMR_DNA-<organism>
                    |- AMR_DNA-<organism>.tab
                    |- changes.txt
                    |- min_software_version.txt
                    |- version.txt
                    |- changes.txt
                    |- taxgroup.tab
                    |- ReferenceGeneCatalog.txt
                    |- ReferenceGeneHierarchy.txt
                    |- amr_targets.fa
               |- YYYY-MM-DD.#
               ...

Note the AMRFinderPlus/data directory contains archived database versions prior to 2019-10-30.1 that are not compatible with more recent releases of the software.

Where AMRFinderPlus/database/latest is a link to the most recent version of the AMRFinderPlus database.

For easier analysis with other tools we provide the reference genes used by AMRFinderPlus along with with additional metadata in the Reference Gene Browser and as a tab-delimited file. The HMMs along with seed alignments are published on our ftp site as well.

Within each database directory there are the following files

• ReferenceGeneCatalog.txt - This is the table we recommend you use if you want data from the AMRFinderPlus database. This is the data we use for the Reference Gene Catalog, and should be the most convenient source of the database.
• ReferenceGeneHierarchy.txt - This is the data behind the Reference Gene Hierarchy, the arrangement of genes, families, and upstream nodes that our curators use to organize and relate the genes and HMMs in the Pathogen Detection Reference Gene Catalog and Pathogen Detection Reference HMM Catalog. This hierarchy drives the gene idenfication and naming algorithm of AMRFinderPlus.
• fam.tab - Tab-delimited file used internally by AMRFinderPlus to define the hierarchical structure behind the AMRFinderPlus database.
• AMRProt - Curated AMR proteins in FASTA format with a custom formatted defline containing metadata used by AMRFinderPlus for each sequence.
• AMR.LIB - A file of curated AMR HMMs with trusted cutoffs.
• AMR_CDS - Coding sequences for each protein in AMRProt. This file is not directly used by AMRFinderPlus, but is included because it is helpful in other analyses.
• AMR_DNA-* - Curated resistance causing nucleotide sequences for point mutation assessment
• AMR_DNA-*.tab - Tab-delimited file of nucleotide point-mutation information
• changes.txt - A human-readable log listing the updates/changes with each release of the database
• min_software_version.txt - To enable software to detect compatible database this is the database format version
• taxgroup.tab - For NCBI Pathogen Detection taxonomic group -> organism mapping (used with --gpipe_org option)
• version.txt - The database version
• amr_targets.fa - The set of targets with reduced sequence redundancy used for the Pathogen Detection guided assembler prcess (SAUTE). These are derived from 'core' genes plus stx only.
• AMRProt-mutation.tab - Tab-delimited file of protein point mutation information
• AMRProt-suppress - Tab-delimited file of -O/--organism argument and protein id to suppress
• AMRProt-susceptible.tab - Tab-delimited file of proteins that are only reported when forms are more divergent

This tab-delimited file is the canonical source of the Pathogen Detection Reference Gene Database. This data is also visible in the Reference Gene Catalog (described here). While this file is not used directly by the AMRFinderPlus software, it should be the place to go if you are interested in the data behind AMRFinderPlus. The other files described here were designed for the use of the AMRFinderPlus software specifically and may change substantially in the future to enable new features of the software.

In this file each row is a protein or point-mutation, and each column is metadata about that protein or point-mutation.

Tab-delimited fields are as follows:

1. allele: If the protein has named alleles (e.g., blaTEM-1) then this field will have the allele symbol for this sequence. If this row is a point mutation then this will contain a point-mutation symbol (e.g., 23S_C2627A) representing a specific point mutation. For point mutations the associated accessions in the other columns will be for the reference gene that does not contain this mutation.
2. gene_family: The gene symbol for this gene family. This is a short abbreviation that is used to refer to this gene. Multiple sequences may share the same symbol, and this symbol will represent multiple very similar gene sequences.
3. whitelisted_taxa: This field is only used for point mutations, and describes the taxon for which this point mutation is relevant. This corresponds to the --organism option to AMRFinderPlus. See also blacklisted_taxa which is used to exclude universal or nearly universal genes.
4. product_name: This is the name of the protein produced by the gene.
5. scope: The data subset to which this row belongs. It will either be 'core' (highly curated, AMR-specific genes and point mutations) or 'plus' (genes related to biocide and stress resistance, general efflux, virulence, or antigenicity).
6. type: Classification of the type of protein/point mutation. This will be AMR, STRESS, or VIRULENCE.
7. subtype: A finer classification for the protein/point mutation if relevant. One of AMR, POINT, ACID, BIOCIDE, ANTIGEN, HEAT, METAL, or VIRULENCE.
8. class: "Class" provides a broad definition of the phenotype affected by the gene or allele, and includes phenotypes such as antimicrobial and stress resistance, virulence, and antigenicity. See class subclass for an enumeration of all classes and subclasses.
9. subclass: Where it is known, "Subclass" provides a more specific definition of the particular antibiotics or classes that are affected by the gene or point mutation (e.g., that are resisted by the gene/allele). While most subclass designations are self-explanatory, a few others have particular meanings. Specifically, "CEPHALOSPORIN" is equivalent to the Lahey 2be definition; "CARBAPENEM" means the protein has carbapenemase activity, but it might or might not confer resistance to other beta-lactams; "QUARTERNARY AMMONIUM" are quaternary ammonium compounds. In addition, stx subtypes (e.g., STX2E) and intimin subtypes (e.g., ALPHA) are defined for Shiga toxin proteins (class of STX1 or STX2) and intimins (class of INTIMIN) respectively. Where the phenotypic information is incomplete, contradictory, or unclear, the "Class" value is used for the "Subclass" value. See [[class subclass]] for an enumeration of all classes and subclasses.
10. refseq_protein_accession: Accession of the RefSeq protein sequence record that defines this entry. In the case of point mutations this will be the accession of the reference sequence which is used to identify the mutation, but is usually a sensitive wild-type allele. In the case of rows identifying proteins, this accession points to the curated canonical sequence for the entry in our database.
11. refseq_nucleotide_accession: Accession of the RefSeq nucleotide sequence that contains the gene coding for the RefSeq protein accession (column 10). Where possible this record contains a small amount of flanking sequence in addition to the coding sequence. Note that this accession may not point to the most common coding sequence for this protein.
12. curated_refseq_start: During curation all sequences and annotations are checked by NCBI curators. On occasion the start site indicated on the GenBank record is missing or deemed incorrect and is set by NCBI curators. In that case this field will be "Yes" and the RefSeq accessions will point to an annotation that is different from that on the GenBank record. In other words the GenBank accession points to a translated protein sequence that is not identical to that included in the database.
13. genbank_protein_accession: The GenBank accession of the protein used to create the RefSeq record. Note that where curated_refseq_start == "Yes" this will not be identical to the curated RefSeq protein sequence.
14. genbank_nucleotide_accession: The GenBank accession of the nucleotide sequence that codes for this protein. This may extend far beyond the coding sequence and include other genes.
15. genbank_strand_orientation: The orientation of the annotation used to create the RefSeq sequence (+/-). If this is "-" then the refseq_nucleotide_accession will point to a sequence that is the reverse complement of the genbank_nucleotide_accession.
16. genbank_cds_start: The coordinate of the start of the coding sequence on the GenBank nucleotide record used to create the RefSeq record.
17. genbank_cds_stop: The coordinate of the end of the coding sequence on the GenBank nucleotide record used to create the RefSeq record.
18. pubmed_reference: PubMed reference associated with this entry.
19. blacklisted_taxa: This protein is common in these taxonomic groups, and does not contribute to acquired phenotype. When AMRFinderPlus is run with an --organism option the presence of this protein will not be reported if the option value matches one of these groups. See also whitelisted_taxa which is used to identify an organism for point-mutation detection.
20. synonyms: Other symbols used to refer to this element/gene.
21. hierarchy_node: For protein sequences, this is the node in the AMR hierarchy to which this protein belongs. The hierarchy itself is described in the ReferenceGeneHierarchy.txt file (and fam.tab). This is not a gene symbol, but a helpful shorthand to keep track of the contents of the node. These are created for the utility and convenience of curators and we make no guarantees that these will persist in content or meaning.
22. db_version: The database version of this entry.

This is the data behind the Reference Gene Hierarchy. The Reference Gene Hierarchy is the arrangement of genes, families, and upstream nodes that our curators use to organize and relate the genes and HMMs in the Pathogen Detection Reference Gene Catalog, Pathogen Detection Reference HMM Catalog. This hierarchy drives the gene idenfication and naming algorithm of AMRFinderPlus.

Tab-delimited fields are as follows:

1. node_id: Unique identifier for this node. This is not a gene symbol, but a helpful shorthand to keep track of the contents of the node. These are created for the utility and convenience of curators and we make no guarantees that these will persist in content or meaning.
2. parent_node_id: Identifier for the parent node.
3. symbol: The gene symbol that will be reported by AMRFinderPlus for a sequence matching at this node and not one more specific.
4. num_prots: The number of proteins assigned to this node or below.
5. hmm_acc: Accession for the HMM assigned to this node (if any)
6. prot_acc: For allelic nodes the accession of the protein at that allele. Multiple proteins can exist at nodes that are non-alleleic, and those can be seen by going to that node in the ReferenceGeneCatalog.txt
7. scope: The data subset to which this row belongs. It will either be 'core' (highly curated, AMR-specific genes and point mutations) or 'plus' (genes related to biocide and stress resistance, general efflux, virulence, or antigenicity).
8. type: Classification of the type of protein/point mutation. This will be AMR, STRESS, or VIRULENCE.
9. subtype: A finer classification for the protein/point mutation if relevant. One of AMR, POINT, ACID, BIOCIDE, ANTIGEN, HEAT, METAL, or VIRULENCE.
10. class: "Class" provides a broad definition of the phenotype affected by the gene or allele, and includes phenotypes such as antimicrobial and stress resistance, virulence, and antigenicity. See class subclass for an enumeration of all classes and subclasses.
11. subclass: Where it is known, "Subclass" provides a more specific definition of the particular antibiotics or classes that are affected by the gene or point mutation (e.g., that are resisted by the gene/allele). While most subclass designations are self-explanatory, a few others have particular meanings. Specifically, "CEPHALOSPORIN" is equivalent to the Lahey 2be definition; "CARBAPENEM" means the protein has carbapenemase activity, but it might or might not confer resistance to other beta-lactams; "QUARTERNARY AMMONIUM" are quaternary ammonium compounds. In addition, stx subtypes (e.g., STX2E) and intimin subtypes (e.g., ALPHA) are defined for Shiga toxin proteins (class of STX1 or STX2) and intimins (class of INTIMIN) respectively. Where the phenotypic information is incomplete, contradictory, or unclear, the "Class" value is used for the "Subclass" value. See class subclass for an enumeration of all classes and subclasses.
12. name: Name for proteins at this node. For reportable nodes this will be reported by AMRFinderPlus if no more specific node is hit above cutoffs.
13. synonyms: Other symbols used for this node is identified in the literature.
14. display_parent: (0/1) Used internally to group notable genes for display.
15. allele: (0/1) 1 if this node is designated as an allele (which means the name/symbol would only be associated with exact matches, any variation and the name/symbol of the parent node would be returned by AMRFinderPlus).
16. db_version: The database version of this entry.

FASTA formatted file containing curated Protein sequences along with the name of the gene/allele that we have assigned them. This FASTA file has a custom formatted defline with additional metadata used by AMRFinderPlus to name elements. Protein sequences are terminated with a "*" character for the stop codon. Fusion proteins have two FASTA entries, one describing the activity of each component. Point mutation references may also be present in mutant forms to aid in blast-based detection of frame-shifts and mutations within one amino-acid of the amino- or carboxy-terminals.

Fields in the defline are separated by '|' characters and are as follows:

1. Protein GI (0 if GI does not exist).
2. Protein accession-version. Mutant point-mutation references indicate the position and name of the mutation with ':' separated values in this field. E.g.: WP_004179093.1:129:pmrB_ALNQLV129del
3. Fusion gene part number (1 or 2 for a fusion gene, 1 if not a fusion gene)
4. Total number of fusion parts (2 for a fusion gene, 1 if not a fusion gene)
5. node_id (see fam.tab) or allele symbol for named alleles
6. The parent_node_id if the protein is an allele, otherwise the same node_id as in field 5
7. Resistance mechanism type
8. Reportability level (0 - do not report, 1 - plus, 2 - core)
9. Subclass
10. Class
11. Protein name with ' ' replaced by '_'

FASTA formatted file containing coding sequences for curated Protein sequences along with the name of the gene/allele that they code for. This FASTA file has a custom formatted defline with additional metadata. Note that there is only one coding sequence included per protein in the database, and the coding sequence included here may not be a common one. This file is currently not used by AMRFinderPlus, but is included because it is often requested. Also note that these sequences include the stop codon. Note that as in the AMRProt file there are duplicated sequences (for fusion genes).

Fields in the defline are separated by '|' characters and are as follows:

1. Protein GI that this sequence codes for
2. Protein accession
3. DNA accession
4. Fusion gene part number (1 or 2 for a fusion gene, 1 if not a fusion gene)
5. Total number of fusion parts (2 for a fusion gene, 1 if not a fusion gene)
6. node_id (see fam.tab) or allele symbol for named alleles
7. The parent node_id if the protein is an allele, otherwise the same node_id as in field 5
8. Resistance mechanism type
9. Protein name with ' ' replaced by '_'

HMM library in HMMER3 ASCII text format. Each model in the AMR.LIB has a format similar to the HMMs of Pfam (see http://pfam.xfam.org ).

For a general description of the file format see the HMMER User's guide at: http://hmmer.org/documentation.html.

Some notes on how we use the descriptive fields in the AMR.LIB file:

• DESC The name applied to sequences hit by this HMM if no more specific HMM or BLAST hit applies. See the section on fam.tab for details about the hierarchical structure behind some of our HMMs.

• NAME Contains an internal node identifier that matches a node_id in the fam.tab file.

• TC The cutoff used by AMRFinderPlus to determine what constitutes a "match" for this HMM.

Cutoffs included in the file (TC, GA, NC): The TC field is the only one of the cutoffs used by and universally curated in our system. HMMs were not used to generate libraries, so we set GC=TC for convenience. NC was set by manual examination of results for some HMMs, but otherwise NC=TC.

Tab-delimited file used internally by AMRFinderPlus to define the hierarchical gene/protein family structure behind the AMRFinderPlus database.

Fields are separated by tab characters and columns are as follows:

1. node_id: The ID of the family described by this line. This is an internal ID used by AMRFinderPlus, and exposed in ReferenceGeneHierarchy.txt and the Reference Gene Catalog. Note that it is not a gene symbol or accession and is subject to change.
2. parent_node_id: Except for the root for all rows this field should have another line for for the parent where the node_id matches this field. The root has an empty value.
3. Gene symbol: The symbol to be reported for hits at this level
4. HMM identifier: The internal identifier (NAME) used to identify the HMM (if any) that is used at this level in the hierarchy
5. HMM trusted cutoff 1: Trusted cutoff for full_score (minmum match for whole HMM, both cutoffs must be met for a hit)
6. HMM trusted cutoff 2: Trusted cutoff for domain_score (minimum match for a region, both cutoffs must be met for a hit)
7. Blast rule cutoff: Complete target identity: This is overridden by the --ident_min option. "0.00" means default behavior (90%, though there will usually be a curated HMM in those cases). Note that this cutoff is used when >= 90% of the reference protein is covered by the hit
8. Blast rule cutoff: Complete target coverage
9. Blast rule cutoff: Complete reference coverage
10. Blast rule cutoff: Partial identity This is overridden by the --ident_min option. "0.00" means default behavior (90%). Note that this rule is used when <90% of the reference protein is covered by the hit
11. Blast rule cutoff: Partial target coverage
12. Blast rule cutoff: Partial reference coverage
13. Reportable 0/1/2: Whether a hit at this level will be reported as an AMRFinderPlus hit (0 = do not report, 1 = 'plus' gene/family, 2 = AMRFinder 'core' gene/family
14. Family name: The gene name to be reported for hits at this level

Note that the only Blast rule cutoffs used by AMRFinderPlus are fields 7 (Complete target identity) and 10 (Partial identity).

This contains point mutation information. The sequence of the reference allele is in AMRProt.

Fields are separated by tab characters and columns are as follows:

1. taxgroup: The "--organism" option that is associated with this mutation
2. accession_version: The accession of the reference protein
3. mutation_position: The offset position in the reference protein sequence for this mutation. This may differ from the position listed in the mutation symbol where the reference sequence doesn't match the traditional numbering system.
4. mutation_symbol: The "gene symbol" that will be printed when this mutation is found
5. class: The functional "class" for this mutation (see [[Class and subclass|Interpreting-results#class-and-subclass]]) for an explanation
6. subclass: The functional "subclass" for this mutation (see [[Class and subclass|Interpreting-results#class-and-subclass]]) for an explanation
7. mutated_protein_name: The contents of the "Sequence name" column when this mutation is found

FASTA formatted file containing DNA reference sequences for the point mutations in non-protein-coding genes or regions. This FASTA file has a custom formatted defline with additional metadata. This includes the accession for the source of the sequence, some naming information, and is the first column in AMR_DNA-<organism>.tab file. Fields are separated by '@' symbols and are as follows.

1. The nucleotide accession that this sequence was derived from
2. The name of the gene or promoter region for the point mutation, ' ' is replaced with '_'
3. A colon (:) separated field with the gene symbol for the gene and the location within the sequence pointed at by the accession in the first field. See the description below for the accession_version@gene_name:start-stop field in AMR_DNA-<organism>.tab file.

This file contains DNA point-mutation information similar to AMRProt-mutation.tab

Fields other than the first are separated by tab characters and columns are as follows:

1. accession_version@gene_name:start-stop: This is a compound '@ and :' separated field that indicates the reference sequence location within the reference accession
   • accession_version: Accession number of the reference sequence
   • gene_name: The gene name of the reference sequence ('_' is replaced by ' ')
   • start-stop: The start and stop coordinates of the reference sequence on the accession provided
2. mutation_position: The position of the mutation in the reference sequence. This may differ from the position listed in the mutation symbol where the reference sequence doesn't match the traditional numbering system.
3. mutation_symbol: The "gene symbol" that will be printed when this mutation is found
4. class: The functional "class" for this mutation (see [[Class and subclass|Interpreting-results#class-and-subclass]] for an explanation)
5. subclass: The functional "subclass" for this mutation (see [[Class and subclass|Interpreting-results#class-and-subclass]]) for an explanation
6. mutated_gene_name: The contents of the "Sequence name" column when this mutation is found

This file lists proteins that are reported only when a protein in the query sequence aligns, but is more divergent than a cutoff. These proteins are only reported for a single --organism. Used for Streptococcus pneumoniae pbp variants

1. taxgroup: --organism to report these genes
2. gene_symbol: Gene symbol to report
3. accession_version: Accession of the reference protein
4. resistance_cutoff: Only report genes more divergent then this cutoff
5. class: The functional "class" for this gene when reported (see [[Class and subclass|Interpreting-results#class-and-subclass]] for an explanation)
6. subclass: The functional "subclass" for this gene when reported (see Class and subclass for an explanation)
7. resistance_protein_name: Element name for this protein when reported

This file lists organisms and genes to not report because for a given --organism because they do not provide additional information for that organism (e.g., they are nearly universal or don't confer a phenotype in that organism)

1. taxgroup: --organism to suppress these genes
2. protein_accession: Accession of the reference protein
3. protein_gi: gi of the reference protein

This file contains a mapping of "taxgroups" used by NCBI Pathogen Detection (see https://www.ncbi.nlm.nih.gov/pathogens) to --organism option values, and is used when the --gpipe_org option is included on the commandline. This is primarily intended for internal use.

Fields are separated by tab characters and columns are as follows:

1. taxgroup: The "--organism" option value
2. gpipe_taxgroup: The NCBI Pathogen Detection taxonomic group
3. number_of_nucl_ref_genes: A count of the number of nucleotide references included for that taxgroup

We also published a paper detailing the curation methods we use: Feldgarden M, Brover V, Fedorov B, Haft DH, Prasad AB, Klimke W. Curation of the AMRFinderPlus databases: applications, functionality and impact. Microb Genom. 2022 Jun;8(6). doi: 10.1099/mgen.0.000832. PMID: 35675101

Current curation of new genes, alleles, and point mutations is primarily from:

• Literature review by curators
• Submissions to NCBI for allele assignment
• Collaboration with Gen-FS working groups
• Comprehensive Antibiotic Resistance Database (CARD) collaboration
• Direct communication / consultation with domain experts

NCBI maintains a collaboration with CARD to resolve issues and communicate updates and new genes. If you know of a gene or point mutation we're missing please email us at pd-help@ncbi.nlm.nih.gov and if possible include the mutation, reference accession, resistances conferred, and at least one citation to make our curators lives a bit easier.

NCBI continually mines the literature for new reports of AMR proteins. The ResFams collection of AMR HMMs provided important help early in our efforts to develop the AMR protein hierarchy and the AMRFinderPlus tool, but all models were rebuilt with new seed alignment sequences, new alignments, new cutoff scores, and new biocuration. Development continued until all reportable AMR proteins were covered by at least one AMR HMM, and classification by HMM was sufficiently specific.

NCBI is also responsible for the assignment of new beta lactamase alleles for certain families. Once new alleles are released, then they are immediately incorporated into AMRFinderPlus. See this page for more information:

https://www.ncbi.nlm.nih.gov/pathogens/submit-beta-lactamase/

"Plus" proteins were covered based on requests from our collaborators at public health agencies, and most are identified using curated blast cutoffs. Many of the "plus" proteins come from classes of genes that are less intensely studied, and so you may be less confident of their impact on phenotype.

To build the AMRFinderPlus HMM and Protein collection, NCBI first assembled a comprehensive collection of acquired (and intrinsic) anti-microbial resistance proteins. Sources, published or collaborative, including

• the Lahey Clinic compilation of beta-lactamase sequences (http://www.lahey.org/studies/ and personal communications from Dr. George Jacoby and Karen Bush)
• the Pasteur Institute collection of beta-lactamase sequences
• ResFinder
• Comprehensive Antibiotic Resistance Database (CARD)
• the RAC and Integrall collections of AMR proteins found in integrons
• the Center for Veterinary Medicine
• Marilyn Roberts personal communications
• "Oxford" - Derrick Crook personal communications

Inputs from all of the above resources were then manually curated, and assembled into a hierarchical database with manually created and curated HMMs, blast rules, gene names, and gene symbols. See the Reference Gene Catalog, Reference Gene Hierarchy and Reference HMM Catalog for web interfaces to view the curated results.

This collection covers AMR resistance proteins only if they confer resistance in bacteria. Alleles for AMR proteins are included in the data set only if they are naturally-occurring. The antibiotic affected by the AMR protein does not need to be used clinically as an antibiotic in human patients. Our collection includes proteins that contribute to resistance quaternary ammonium compounds (which are not antibiotics, strictly speaking) and to antibiotics whose use is restricted to agricultural or veterinary applications, e.g. olaquindox. Source databases we drew from (e.g. CARD) contained a large number of intrinsic proteins that contribute weakly to resistance (loss-of-function mutants show increased susceptibility), and that may contribute more strongly after mutational events increase their expression. We avoided including most such proteins in the release, since flagging such proteins makes the reports on AMR proteins identified far more difficult to read and understand.

As part of the 'core' database, known resistance-associated point mutations are included for some organisms, but others remain not-covered. See --organism option docuemntation for more which organism groups are covered. Files provided will not help much in finding sources of resistance in Mycobacterium tuberculosis, where nearly all the recent increase in antimicrobial resistance is attributable to mutational changes that are not included in the database. We are actively curating and increasing coverage of point-mutations.

At the request of our collaborators we have added an expanded set of genes that are of interest in pathogens. This set includes stress response (biocide, metal, and heat resistance), virulence factors, some antigens, and porins. These "plus" proteins have been primarily been added to the database with curated blast cutoffs, and are generally identified by blast searches. Some of these may not be acquired genes or mutations, but may be intrinsic in some organisms.

As we collected AMR proteins from various sources, we examined them in multiple sequence alignments to determine whether we judged any to have structural problems, including truncations, frameshifts, or incorrect start sites. For aminoglycoside-modifying enzymes found in integrons, in particular, the most appropriate start site to choose often is unclear. We chose start sites such that regions of homology across a family were preserved, but regions clearly derived from the sites at which the genes integrated were removed. Note that this is separate from the alignment trimming step below.

All HMMs used by AMRFinderPlus can be found at the Reference HMM Catalog, in the database file AMR.LIB in HMMER3 format, and with seed alignments and some additional data on our NCBIfam ftp site. This library is a special subset from a larger collection of protein profile HMMs being constructed at NCBI, and referred to on the whole as NCBIfams. NCBIfams is a collection of protein family hidden Markov models (HMMs) for improving bacterial genome annotation. A paper by Haft et al. 2018 provides additional information about NCBIfams, which is part of NCBI's Reference Sequence (RefSeq) project.

The AMRFinderPlus HMM library consists of a subset of models designed for detecting and classifying candidate AMR proteins. Each model was built from a manually reviewed and trimmed multiple sequence alignment, called the seed alignment. Regions of sequence that appeared extraneous, as from incorrect prediction of an upstream start site, were removed by trimming. Sequences that appeared non-representative, such as those with frameshift mutations or truncations, were removed. Models were built using the HMMER3 package (http://hmmer.org; Eddy, 2011). For some HMMs, the seed alignment may consist of a single sequence.

Three types of cutoffs are provided, for compatibility: trusted_cutoff (TC), gathering threshold (GA), and noise_cutoff (NC). In this library, TC and GA are always set to be identical. If NC is set to be the same as TC, this indicates that the noise cutoff has not yet been manually reviewed. Because all three types of cutoff are provided, searches may be performed using any of the HMMER package switches, --cut_tc, --cut_ga, or --cut_nc. AMRFinderPlus uses --cut_tc.

For virulence and stress resistance genes and AMR point mutations AMRFinderPlus curation is ongoing and organism specific data has not have been curated for all taxa. The Curated organisms table indicates which organisms and element types NCBI Pathogen Detection curators believe they have good coverage in the database for, and that AMRFinderPlus with the latest database version should have good coverage to identify known phenotypically important genes or point mutations.

AMRFinderPlus treats all families and alleles of AMR proteins as nodes in hierarchical tree. Below is an example of the parent-child relationships for a beta-lactamase allele and the broader families that contain it.

bla - beta-lactamase
  bla-A - class A beta-lactamase
    bla-A_carba - carbapenem-hydrolyzing class A beta-lactamase
      blaKPC - KPC family carbapenem-hydrolyzing class A beta-lactamase
        blaKPC-2 - carbapenem-hydrolyzing class A beta-lactamase KPC-2 (allele)

Evidence used to annotate proteins relies on HMM and BLAST. AMRFinderPlus will assign the most specific name it can find that is justified by the evidence. In the example shown above, the bottom level represents an allele. Assigning an AMR protein to a specific allele requires a 100% identity BLAST match. Each higher level node may have an associated HMM, with defined cutoff scores that make assignment to that family deterministic. Not every (non-allele) node has its own HMM, but every AMR protein is covered by at least one HMM somewhere above it in the hierarchy.

Note that some nodes in the hierarchy group together child families that are not necessarily related by homology. Such nodes will always lack an HMM. For example, the family bla ("beta-lactamase") contains child families bla-A ("class A beta-lactamase") and "metallo-beta-lactamase", which are similar in function but lack any sequence homology to each other. We now have a Reference Gene Hierarchy browser available with links to proteins in the Reference Gene Catalog and Reference HMM Catalog.