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

See the Running AMRFinderPlus page for details on what the output fields are.

There are several columns that can help inform your understanding of function. See An example report for an example of the possible values for the different columns described below. To briefly summarize, the "Method" column combined with the % identity indicates how a hit was identified, and whether the protein is a partial, an exact match to a known sequence, etc. This should give an indication of how confident you should be that this is a full-length functional gene. The "core" vs. "plus" column should help you to determine the level of curation that went into the inclusion of that gene in the database. The Sequence name, Type, Subtype, Class, and Subclass provides an estimation of the category and function for the gene.

AMRFinderPlus splits the database into two subsets. By default it only returns genes in the 'core' subset.

• "Core": this subset includes highly curated, AMR-specific genes and proteins from the Bacterial Antimicrobial Resistance Reference Gene Database (BioProject PRJNA313047), plus point mutations. The sources of input for this curated database include: 1) allele assignments, 2) exchanges with other external curated resources, 3) reports of novel antimicrobial resistance proteins in the literature.
• "Plus": this subset includes genes related to biocide and stress resistance, general efflux, virulence, or antigenicity, or AMR genes whose presence/absence are unlikely to affect phenotype and/or is highly uncertain. These genes are only shown if the --plus option is used.

Important information about how likely the hit is to be functional and/or how novel a hit is can be found in the "method" column. A description of possible values is listed on the Running AMRFinderPlus page. Note that for BLAST-based "Methods" there is a suffix of either 'X' or 'P' that indicates whether it was found using translated genomic sequence or protein sequence.

The methods that can apply to either nucleotide or protein sequence have a suffix appended to designate which type of search was used for that line in the report. The suffix indicates whether the line was identified in a protein (P) or nucleotide translated (X) input file. Nucleotide BLAST (BLASTN) is used to identify nucleotide point mutations and POINTN is used as the Method for those hits.

Note as of version 3.10.11 BLASTX / POINTX / EXACTX results may actually come from tblastn. Tblastn is used instead of blastx to increase speed in some cases.

These mean that we have an amino-acid sequence identical protein in the AMRFinderPlus database. The ALLELE method is reserved for genes that have an allele numbering scheme, so the "Gene symbol" for an ALLELE hit is really an allele symbol specific for that exact sequence. EXACT matches have an identical amino-acid sequence in the database, but the family does not have alleles so the same gene symbol may be shared with other closely related sequences. Note that if you run a combined search (including -p, -n, and -g options) and you get an ALLELEX or EXACTX result that means that the annotation did not contain the protein.

"BLAST" hits are hits that are < 100% identical to a database protein, but at coverage > 90%. The percent identity cutoff is, by default, 90%, but may be higher or lower if it was manually curated. Manual curation of BLAST parameters usually occurrs for 'plus' genes where no HMM and cutoff have been curated.

Hits < 90% of the length of the database protein are called either "PARTIAL" if the hit is internal to a contig or "PARTIAL_CONTIG_END" if the gene could have been broken by a contig boundary. "PARTIAL_CONTIG_END" require that the protein alignment ends within two nucleotides of the end of the contig with an orientation such that the end of the protein would extend off the end of the contig. Because assemblers sometimes split genes over multiple contigs, genes that are PARTIAL_CONTIG_END are often full-length in reality. Note that searches using only protein FASTA files (-p <protein_fasta) do use contig location and so all < 90% coverage hits will just be called PARTIAL.

Note that by default AMRFinderPlus does not identify gene fragments that cover < 50% of the reference, so alignments from genes marked PARTIAL cover between 50% and 90% of the reference. The default minimum coverage can be changed by using the --coverage_min option (see Usage (syntax/options) for more information).

These are genes that when translated from genomic sequence have a stop codon before the end of the database protein. These are less likely to be functional, and can only be assessed if the -n <nucleotide_fasta> option is used.

HMM-only hits happen when the criteria for a BLAST hit is not met and an HMM match is above the curated cutoff for an HMM that has been created for that gene or gene family. These will usually be distant relatives of known gene families and may be candidates for a new gene family. Occasionally, partial proteins that have diverged enough from known database proteins to not meet the BLAST cutoffs will show up as HMM-only hits.

Point mutation detection is only enabled when an --organism option is included for a particular taxonomic group. AMRFinderPlus does not report all point mutations in those genes, only ones that have been found in papers describing their functional relevance. Thus the absence of a POINTP, POINTX, or POINTN result does not mean that there are no functional point mutations, only that AMRFinderPlus was unable to find a gene with known point mutations. This could be either because the gene does not exist in the assembly, or the functional point mutation was not one included in the database. See The Pathogen Detection Reference Gene Browser for a list of known point mutations. The --mutation_all option will create a file with all genotype calls made by AMRFinderPlus. With that output you can assess whether or not a specific gene was identified in the query sequence.

These fields describe the classification of the AMRFinderPlus gene. "Element type" is split into 3 categories, AMR, STRESS, or VIRULENCE. "Element subtype" is a duplicate of "Element type" unless a more specific category has been defined.

Element type	Element subtype	Description
AMR	AMR	Antimicrobial resistance gene
AMR	AMR-SUSCEPTIBLE	Not used in AMRFinderPlus output, but in the Reference Gene Catalog (see note below)
AMR	POINT	Known point mutation associated with antimicrobial resistance
VIRULENCE	VIRULENCE	Virulence gene
VIRULENCE	ANTIGEN	Gene codes for a known antigen (these are often used for typing)
STRESS	ACID	Acid resistance gene
STRESS	BIOCIDE	Biocide resistance gene
STRESS	HEAT	Heat resistance gene
STRESS	METAL	Metal resistance gene

This is used for Pencillin-resistant Streptococcus pneumoniae pbp alleles, which are defined as those alleles which are below a protein identity threshold to a set of pbp1a, pbp2b, or pbp2x reference alleles: the thresholds are 99%, 99%, and 98% for pbp1a, pbp2b, and pbp2x respectively. The susceptible reference alleles can be found in NCBI's Reference Gene Catalog where they have subtype AMR-SUSCEPTIBLE to indicate that the reference is the susceptible version. These alleles may also confer resistance to other beta-lactam antibiotics. If you are using AMRFinderPlus with Streptococcus pneumoniae pbp genes you may need to analyze further as necessary. Subtype AMR-SUSCEPTIBLE will not appear in AMRFinderPlus output because only putatively resistant forms will be identified.

These are classifications of the effect of the gene on resistance to various stresses. Be aware that AMRFinderPlus identifies genetic elements, but genotype and phenotype may not correspond (see also note regarding genotype and phenotype). The class and subclass fields are also used to add typing information in the cases of stx and intimins. These fields are blank for many "plus" genes.

For a list of all possible combinations of class and subclass see class-subclass. For AMR genes "Class" can be thought of as drug class, and 'Subclass' contains a more specific drug designations where known. 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 quarternary ammonium compounds. Where the phenotypic information is incomplete, contradictory, or unclear, the "Class" value is used for the "Subclass" value.

For some antigen and virulence genes that are often referred to by type, specific type information is included here as follows:

For eae (Intimin) genes the "Class" is INTIMIN, while the subclass contains the intimin type (ALPHA, BETA, EPSILON, GAMMA, IOTA, LAMBDA, or RHO).

For stx (Siga toxin) genes the possibilities for "Class" and "Subclass" provide typing information as follows:

Class	Subclass
STX1	STX1A
STX1	STX1B
STX1	STX1C
STX1	STX1D
STX2	STX2A
STX2	STX2B
STX2	STX2C
STX2	STX2D
STX2	STX2E
STX2	STX2F
STX2	STX2G

Handling of fusion genes is still under active development. Currently they are reported as two lines, one for each portion of the fusion. Gene symbol, Protein name, Name of closest protein, HMM id, and HMM description are reported with respect to the individual elements of the fusion. This behavior is subject to change.

AMRFinderPlus has no explicit detection of frame shifts. Frame shift mutations in nucleotide inputs will usually result in genes detected with the "method" INTERNAL_STOP or PARTIAL, but that will depend on the nature of the frame shift and the sequence following it.

File format checking is rudimentary. Software behavior with incorrect input files is not defined, and error messages may be cryptic. Email us if you have questions or issues and we'll be happy to help and use your input to improve our error checking and messages.

If you find bugs or have other questions/comments please email us at pd-help@ncbi.nlm.nih.gov.