Introduction to SWISS-MODEL Repository (SMR)

The SWISS-MODEL Repository (SMR) is a database of 3D protein structure models generated by the SWISS-MODEL homology modelling pipeline. It provides access to an up-to-date collection of annotated 3D protein models generated by automated homology modelling for relevant model organisms and experimental structure information for all sequences in UniProtKB. Regular updates ensure that target coverage is complete, that models are built using the most recent sequence and template structure databases, and that improvements in the underlying modelling pipeline are fully utilised. It also allows users to assess the quality of the models using the latest QMEAN results. If a sequence has not been modelled, the user can build models interactively via the SWISS-MODEL workspace. Annotation of models with functional information and cross-linking with other databases, such as UniProtKB, facilitates the navigation between protein sequence and structure resources.

If you are using models from the SWISS-MODEL Repository, please cite the following articles:

  • Bienert S, Waterhouse A, de Beer TAP, Tauriello G, Studer G, Bordoli L, Schwede T
    The SWISS-MODEL Repository - new features and functionality.
  • Waterhouse A, Bertoni M, Bienert S, Studer G, Tauriello G, Gumienny R, Heer FT, de Beer TAP, Rempfer C, Bordoli L, Lepore R, Schwede T
    SWISS-MODEL: homology modelling of protein structures and complexes.

Data content and update

SMR currently provides regular updates of models for the Swiss-Prot section of UniProtKB and for the reference proteomes of the following core species: H. sapiens, M. musculus, C. elegans, E. coli, A. thaliana, D. melanogaster, S. cerevisiae, S. pombe, C. vibrioides, M. tuberculosis, P. aeruginosa, S. aureus, and P. falciparum.

For each core species proteome, SMR provides a dedicated summary page with information on the target proteome, statistics on model coverage as well as the analysis of structural coverage of the proteome over time. The full set of model data for each core species is downloadable from the main entry page as well as the summary page. Models for non-core sequences are added on demand by interactive user request.

A continuous release mechanism ensures that the models are regularly updated based on the latest template information and the latest version of the SWISS-MODEL modelling pipeline. The current version of the SMR features a three-tiered policy for updating: monthly, weekly, and on-demand.

Monthly updates are initiated when a new version of UniProtKB is released, i.e. the amino acid sequences of UniProtKB entries may have changed or the composition of the core species proteomes may have been updated.

Weekly updates ensure that the latest PDB release is readily integrated in the SWISS MODEL Template Library (SMTL), and models are built using the latest template information.

On-demand updates are meant for sequences that are not part of the core species data set. When accessing an SMR entry for which no models have been built, or the existing models are not based on current template information, the user is presented with two options: (i) requesting an automated update of the SMR entry (ii) starting an interactive modelling job on the SWISS-MODEL Workspace.

Note: Only models of high quality are imported into SMR! The QMEANDisCo global score of a model must be greater than 0.5.


Accessing data

The SMR web interface provides several entry points for accessing the data. The simplest way to directly access an entry is via its UniProtKB accession code. Alternatively, free text search on protein names, functional description, and organisms allows selecting entries from a list of results matching the search keywords.

The main page offers a general summary about the repository content and, for each of the core species, the following information: total number of canonical sequences in the reference proteome (according to UniProtKB), number of sequences for which at least one model is present, total number of models for that species, and a sequence coverage plot showing the number of sequences that have models or structures in a certain sequence coverage range (>80%, between 60% and 80%, etc.). A colour gradient, from dark-green to red, specifies different coverage values, from high to low respectively. The size of each box is proportional to the number of target sequences within a given range of coverage (e.g. for E. Coli, more than 60% of the proteins are covered by more than 80% over their sequence length by structures and models).

A download button allows to download all the metadata about models and structures of a given core species (further details are provided in the help file in the downloaded folder as well as in the "SMR API" section below).


Model species page

By clicking on the core species names the user can access a dedicated summary page, with information on the target proteome, statistics on model coverage as well as the evolution of structural coverage of the proteome over time. The full set of model data for each core species is downloadable from the main entry page as well as the summary page.


SWISS-MODEL Repository entry page

Individual entries can be searched via UniProtKB accession codes or using a free text search on protein names, functional description and/or organisms. In the latter case the user is provided with a list of entries matching the search keywords and access the information of each entry by selecting from the list.

The entry view page for a specific entry provides four general sections: (i) a graphical representation of the target coverage by models and experimental structures, (ii) a SMR entry summary, (iii) a 3D structure visualisation of the entry, and (iv) the target–template sequence alignment. Each of these sections is described below.

Structural coverage

The amino acid sequence of the target is indicated with a scale. Information of the structural coverage is indicated below the sequence, where solid green segments indicate available experimentally determined structures and dashed blue outlines indicate homology models.

Sequence features from UniProt are displayed below the sequence and include InterPro domains, PFAM domains, variants, transmembrane regions, disulphide bonds, nucleotide binding regions, signal peptides, and active site residues.

As the list of available structures for an entry can be very large, additional structures may be found in the lists below the coverage display.

When available, models from the AlphaFold DB are appended to the list of available experimental structures and models. If you use an AlphaFold prediction in your work, please cite the AlphaFold methods paper: Jumper, J et al. Highly accurate protein structure prediction with AlphaFold. Nature (2021).


Entry summary

The entry summary provides general information about the target protein, including the name of the protein and links to related information in UniProtKB, InterPro, and STRING databases.

For each selected model, the following information is provided: the PDB ID of the structure used as a template for modelling, the SMTL version used (i.e. the date when the SMTL was last updated), the target–template sequence identity and the model confidence (allowing for the identification of unreliable or poorly modelled regions as described here).

In case the user prefers to build a model using a different template structure than the one selected automatically, e.g. in a different conformation state or with a specific ligand, an "Interactive Modelling" button is provided with a link to the SWISS-MODEL workspace. Here the user can start a new interactive modelling session with the target sequence preloaded.


3D structure visualisation

This section provides an in-page visualisation of the currently selected model/structure. The graphical view has been implemented using PV, an interactive JavaScript/WebGL based 3D structure viewer, and NGL, a web application for molecular visualisation. All sections in the SMR entry page are linked to the 3D structure viewer. Any selected feature from the entry summary or the sequence alignment can be visualised on the structure and vice versa (i.e. a residue selected in the structure is highlight in the alignment).


Target–template sequence alignment

Details regarding the alignment between the target and the template are provided. The configuration button allows applying various colouring schemes to the sequence and the 3D model shown. Amongst others, the user can colour the alignment and the structure by amino acid properties, model confidence, chain or secondary structure. An option is also provided to export the alignment in any of the following formats: FASTA, Clustal and PNG.


Custom parameters

The selection of the structure displayed initially, as well as several aspects of the initial display can be customized with URL parameters. Many of these can be used in combination.

An example of custom URL with combined parameters is the following:

https://swissmodel.expasy.org/repository/uniprot/P0DTD1?template=6m71.1.A&range=4423-5311

  • https://swissmodel.expasy.org/repository/uniprot/P0DTD1?template=6m71.1.A

    Select the homology model based on chain A of entry 6m71.1 of the SWISS MODEL Template Library (SMTL). If a structure from the PDB is available, the 4 alphanumeric characters of the PDB ID may be entered as follows:

    https://swissmodel.expasy.org/repository/uniprot/P0DTD1?template=6w9c

  • https://swissmodel.expasy.org/repository/uniprot/P0DTD1?range=1024-1194

    https://swissmodel.expasy.org/repository/uniprot/P0DTD1?from=1024&to=1194

    Select a homology model or experimental structure that covers residues 1024-1194 of the target sequence.

  • https://swissmodel.expasy.org/repository/uniprot/P07900?csm=969F65FCC0BC86FD

    The CRC64 checksum of the protein sequence. This can be used to ensure the sequence for the given UniProtKB entry matches our record.

  • https://swissmodel.expasy.org/repository/uniprot/P0DTD1?id=PRO_0000449630

    For polyproteins which cleave into smaller mature proteins, the Protein ID for the given UniProtKB entry will be displayed.

  • https://swissmodel.expasy.org/repository/uniprot/P0DTD1?annot=3

    If annotations have been uploaded and a project is active, display the annotation in row 3 (note that the numbering starts from 0). This is equivalent to clicking on the annotation after the page loads.

Model quality

Each SWISS-MODEL model in SMR is evaluated by QMEANDisCo (Studer et al.) to provide model quality measures on a per-residue basis as well as a global scale. The graphs provide plots of the estimated local quality of each part of the model. In the main overview graph, colours are used to provide quality information at a glance with blue indicating good and red indicating bad quality scores. Per-residue quality values are also provided in the B-factor column of the PDB file of the model. More information can be found in the SWISS-MODEL help section.

Models from the AlphaFold DB are appended to the available structures / models if available. For these models we use the confidence values provided by AlphaFold (pLDDT) rescaled to be between 0 and 1. Since both pLDDT and QMEANDisCo are trained to predict lDDT (Cα-only for pLDDT and all-atom for QMEANDisCo) and are displayed in the same range, they should be considered comparable.


On-demand update of individual entries

If no structure or model is available for a given UniProtKB entry or it is possible that new templates were added since the last modelling attempt, the user can request for models to be rebuilt or start an interactive modelling job using the SWISS-MODEL Workspace.

Please notice that it may take up to 15 minutes before a model is added to the SMR and that only models of high quality are imported.


SMR API full details available in this OpenAPI document.

SWISS-MODEL Repository is continuously updated to provide the user with with the best available model/experimental structure for a given UniProtKB AC. Due to this continuous update, there are no fixed URLs to specific versions of models, and indeed, models may no longer exist if a better template becomes available in the SWISS-MODEL Template Library.

To find out what is currently available in SMR, first get a list of available models and experimental structures in JSON format (here UniProtKB AC P07900 is used as an example).

  • https://swissmodel.expasy.org/repository/uniprot/P07900.json

    Returns a JSON object, {"result":"structures":[]} is a list of homology models and experimental structures for UniProtKB entry P07900

  • https://swissmodel.expasy.org/repository/uniprot/P07900.json?provider=swissmodel

    Returns a JSON object, {"result":"structures":[]} is a list of homology models (provider=swissmodel) for UniProtKB entry P07900

  • https://swissmodel.expasy.org/repository/uniprot/P07900.json?template=5fwl.1.A

    Returns a JSON object, {"result":"structures":[]} is a list of homology models based on the template 5fw1.1 and chain A. When including experimental structures only the PDB code is required, eg 5fw1

  • https://swissmodel.expasy.org/repository/uniprot/P07900.json?range=15-250

    Returns a JSON object, {"result":"structures":[]} is a list of homology models or experimental structures that have residues in range 15–250 of the target sequence

From the list of available models, the user can select a specific model or structure and fetch the coordinates using the URL provided in the "coordinates" field.

As previously stated, the continuous update of SMR means a specific model may not exist anymore, so no specific model ID is available. However, for a given entry, the API will return the best model/structure available. The best model is the the one with the highest quality based on QMEANDisCo global score. The "sort" parameter allows to sort the models by sequence identity (seqid) or sequence similarity (seqsim) to template. Experimental structures can be sorted based on stoichiometry, range, experimental method, and resolution. If no provider is specified, experimental structures will be returned before homology models.

An example of combined URL is the following:

https://swissmodel.expasy.org/repository/uniprot/P07900.pdb?from=14&to=223&template=2k5b&provider=pdb

If no model is found for your requested coordinates, a 404 Not Found error will be raised.

  • https://swissmodel.expasy.org/repository/uniprot/P07900.pdb

    Returns the best homology model or experimental structure in PDB format.

  • https://swissmodel.expasy.org/repository/uniprot/P07900.pdb?sort=seqsim

    Returns the best homology model or experimental structure in PDB format sorted by sequence similarity.

  • https://swissmodel.expasy.org/repository/uniprot/P07900.pdb?sort=seqid

    Returns the best homology model or experimental structure in PDB format sorted by sequence identity.

  • https://swissmodel.expasy.org/repository/uniprot/P07900.pdb?provider=swissmodel

    Returns the best homology model from the given provider in PDB format.

  • https://swissmodel.expasy.org/repository/uniprot/P07900.pdb?template=5fwl.1.A

    Returns the best homology model or experimental structure based on the given template in PDB format.

  • https://swissmodel.expasy.org/repository/uniprot/P07900.pdb?range=15-250

    Returns the best homology model or experimental structure which overlaps the given range.