Skip to main content

Streptococcus pyogenes in silico EMM typer

Project description

emmtyper - Emm Automatic Isolate Labeller

CircleCI Coverage Status PyPI - Implementation PyPI - Python Version PyPI PyPI - License PyPI - Wheel Conda PyPI - Downloads PyPI - Status GitHub issues

Table of Content


emmtyper is a command line tool for emm-typing of Streptococcus pyogenes using a de novo or complete assembly.

By default, we use the U.S. Centers for Disease Control and Prevention trimmed emm subtype database, which can be found here. The database is curated by Dr. Velusamy Srinivasan. We take this opportunity to thank Dr. Srinivasan for his work.

Inner workings

The difficulty in performing M-typing is that there is a single gene of interest (emm), but two other homologue genes (enn and mrp), often referred to as emm-like. The homologue genes may or may not occur in the isolate of interest. When performing emm-typing from an assembly, we can distinguish betweeen one or more clusters of matches on the contigs. The best match for each of the clusters identified is then parsed from the BLAST results. Where possible, we try to distinguish between matches to the emm gene, and matches to one of the emm-like genes.

Possible arrangments:



        mrp                 emm                 enn


           emm                 enn



  • blastn ≥ 2.6 (tested on 2.9)
  • isPcr
  • python ≥ 3.6



brew install python blast ispcr
pip3 install emmtyper
emmtyper --help


conda install -c conda-forge -c bioconda -c defaults emmtyper


emmtyper has 2 workflows: directly BLASTing the contigs against the DB, or using isPcr to generate an in silico PCR product that is then BLASTed against the DB. The BLAST results go through emmtyper's business logic to distinguish between emm and emm-like alleles and derive the isoolate M-type.

The basic usage of emmtyper is in the form of:

emmtyper [options] contig1 contig2 ... contigN

All the available options can be inspected with emmtyper --help. Options passed on to blast are tagged with [BLAST], and those for isPcr are tagged with [isPcr].

Usage: emmtyper [OPTIONS] [FASTA]...

  Welcome to emmtyper.


  emmtyper *.fasta

  --version                       Show the version and exit.
  -w, --workflow [blast|pcr]      Choose workflow  [default: blast]
  -d, --blast_db TEXT             Path to EMM BLAST DB  [default:
  -k, --keep                      Keep BLAST and isPcr output files.
                                  [default: False]
  -d, --cluster-distance INTEGER  Distance between cluster of matches to
                                  consider as different clusters.  [default:
  -o, --output TEXT               Output stream. Path to file for output to a
                                  file.  [default: stdout]
  -f, --output-format [short|verbose|visual]
                                  Output format.
  --dust [yes|no|level window linker]
                                  [BLAST] Filter query sequence with DUST.
                                  [default: no]
  --percent-identity INTEGER      [BLAST] Minimal percent identity of
                                  sequence.  [default: 95]
  --culling-limit INTEGER         [BLAST] Total hits to return in a position.
                                  [default: 5]
  --mismatch INTEGER              [BLAST] Threshold for number of mismatch to
                                  allow in BLAST hit.  [default: 4]
  --align-diff INTEGER            [BLAST] Threshold for difference between
                                  alignment length and subject length in BLAST
                                  hit.  [default: 5]
  --gap INTEGER                   [BLAST] Threshold gap to allow in BLAST hit.
                                  [default: 2]
  --blast-path TEXT               [BLAST] Specify full path to blastn
  --primer-db TEXT                [isPcr] PCR primer. Text file with 3
                                  columns: Name, Forward Primer, Reverse
                                  Primer.  [default:
  --min-perfect INTEGER           [isPcr] Minimum size of perfect match at 3\'
                                  primer end.  [default: 15]
  --min-good INTEGER              [isPcr] Minimum size where there must be 2
                                  matches for each mismatch.  [default: 15]
  --max-size INTEGER              [isPcr] Maximum size of PCR product.
                                  [default: 2000]
  --ispcr-path TEXT               [isPcr] Specify full path to isPcr
  --help                          Show this message and exit.

Most of these options are self explanatory. The two expections are:

  1. --workflow: choose between a blast only workflow, or a in silico PCR followed by blast workflow. See below for more information.
  2. --clust_distance defines the minimum distance between clusters of matched sequences on the contigs to generate separate emm-type calls for each clusters. Clusters of matches that are within the minimum clust-distance are treated as a single location match.
  3. --output_type demonstrated below.

Example Commands

# basic call using the blast workflow for a single contig file
emmtyper isolate1.fa
# basic call using the pcr workflow for all the .fa files in a folder
emmtyper -w pcr *.fa
# basic call changing some of the options for blast
emmtyper --keep --culling_limit 10 --align_diff 10 *.fa
# call using the pcr workflow changing some of the isPcr options and
# using the visual output format
emmtyper -w pcr --output-format visual --max-size 2000 --mismatch 5 *.fa

Result Format

Short format

emmtyper has three different result formats: short, verbose, and visual.

emmtyper by default produces the short version. This consists of five values in tab-separated format printed to stdout.

The values are:

  • Isolate name
  • Number of clusters: should be between 1 and 3, larger values could indicate contamination
  • Predicted emm-type
  • Possible emm-like alleles (semi-colon separated list)
  • EMM cluster: Functional grouping of EMM types into 48 clusters

Verbose format

The verbose result returns:

  • Isolate name
  • Number of BLAST hits
  • Number of clusters: should be between 1 and 3, larger values could indicate contamination
  • Predicted emm-type
  • Position(s) emm-like alleles in the assembly
  • Possible emm-like alleles (semi-colon separated list)
  • emm-like position(s) in assembly
  • EMM cluster: Functional grouping of EMM types into 48 clusters

The positions in the assembly are presented in the following format <contig_number>:<position_in_contig>.

Visual format

The visual result returns an ASCII map of the emm and, if found any emm-alleles, in the genome. Alleles on a single contig are separated by "-", with each "-" representing 500bp. Alleles found on different contigs are separated with tab.


The alleles can be tagged with a suffix character to indicate different possibilities:

Tag Description Additional Information
* Suspect emm-like Allele flagged in the CDC database as possibly emm-like
~ Imperfect score Match score below 100%

Example outputs

Example for all result format:

Short format:

Isolate1	1	EMM65.0		E6
Isolate2	3	EMM4.0	EMM236.3*;EMM156.0*	E1
Isolate3	2	EMM52.1	EMM134.2*	D4

Verbose format:

Isolate1	6	1	EMM65.0	5:82168	E6
Isolate2	8	3	EMM4.0	2:104111	EMM236.3*;EMM156.0*	2:102762;2:105504	E1
Isolate3	5	2	EMM52.1	14:10502	EMM134.2*	5:913	D4

Visual format:

Isolate1	EMM65.0
Isolate2	EMM156.0*--EMM4.0--EMM236.3*
Isolate3	EMM52.1	EMM134.2*


If you are not sure which pipeline to choose from, we recommend using blast first. The blast workflow is fast and works well with assemblies. You can then use the pcr mode if you wish to perform some troubleshooting.

For example, the pcr workflow might be useful when troubleshooting isolates for which emmtyper has reported more than 3 clusteres and/or too many alleles.

An important thing to note is that not all emm-like alleles can be identified by using by PCR typing. The pcr workflow can be used to test which hits would be returned if carrying out conventional M-typing using PCR. However, the workflow is not foolproof, as in silico PCR will fail when one or both primers do not align in the same contig (i.e., the allele is broken across two or more contigs) or there are mutations in the primer sites. In the former case, this might be an indication of poor sequence coverage or contamination.

Validation data

We compared emmtyper against Sanger sequencing data and PHE's tool emm-typing-tool.

You can check out the validation comparison go to out binder:



  • Andre Tan
  • Torsten Seemann
  • Jake Lacey
  • Mark Davies
  • Liam Mcintyre
  • Hannah Frost
  • Deborah Williamson
  • Anders Gonçalves da Silva

The codebase for emmtyper was primarly written by Andre Tan as part of his Master's Degree in Bioinformatics. Torsten Seemann, Deborah Williamson, and Anders Gonçalves da Silva provided supervision and assistance.

Hannah Frost contributed with EMM clustering by suggesting we incorporate it in to the code, and providing the necessary information to do so and test it.

Jake Lacey, Liam Mcintyre, and Mark Davies provided assistance in validating emmtyper.


The code is actively maintained by MDU Bioinformatics Team.

Contact the principal maintainer at andersgs at gmail dot com.


Please post bug reports, questions, suggestions in the Issues section.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

emmtyper-0.2.0.tar.gz (171.5 kB view hashes)

Uploaded source

Built Distribution

emmtyper-0.2.0-py3-none-any.whl (183.4 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page