Interpreting results

Genotype vs. Phenotype

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 by plasmids, correlations of genotype to phenotype are much more easily deciphered, but users are cautioned against over-interpretation.

Summary

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.

The "Scope" column and the --plus option

By default AMRFinderPlus presents a list of genes of use to those interested primarily in antimicrobial resistance prediction or epidemiology. These genes have a "Scope" value of "core", and are the set of genes and mutations that we find to have evidence of phenotypic effect in wildtype backgrounds. In a few cases the curators may also include genes that have certain functional alleles, or are of particuar public health interest.

The "Method" column

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 most 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.

EXACTX, EXACTP, ALLELEX, and ALLELEP

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 "Gene symbol" can describe 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.

BLASTX and BLASTP

"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.

PARTIALX, PARTIALP, PARTIAL_CONTIG_ENDX, and PARTIAL_CONTIG_ENDP

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. 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 not know about contig boundaries and so all <90% coverage hits will just be called PARTIAL.

INTERNAL_STOP

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

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 known database proteins to not meet the BLAST cutoffs will show up as HMM-only hits.

POINTX and POINTP

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 funtional relevance. Thus the absence of a POINTP or POINTX 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.

Element Type and Subtype

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	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	BIOCIDE	Biocide resistance gene
STRESS	HEAT	Heat resistance gene
STRESS	METAL	Metal resistance gene

Class and Subclass

These are classifications of the effect of the gene on resistance to specific substences. These fields are also used to add typing information in the cases of stx and intimins. These fields are blank for many "plus" genes.

class-subclass is the list of possible value combinations for the class and subclass fields. For AMR genes "Class" can be thought of as drug class, and 'Subclass' contains a more specific drug designations where known. 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

Known Issues

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 with respect to the individual elements of the fusion. This behavior is subject to change.

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.