Source and file download


The webserver relies on two different pipelines: Profs and ELASPIC.

Profs unifies domain definitions obtained from Pfam and CATH in order to create domain boundaries that are the most consistent with both with the structural fold and the evolutionary conservation of the protein. Profs domain boundaries for the entire Uniprot database can be downloaded here. An R package implementing the Profs pipeline is hosted on bitbucket.

ELASPIC constructs homology models of domains and domain-domain interactions, and uses those models, together with sequential and other features, to predict the energetic impact of a mutation on the stability of a single domain or the affinity between two domains. A more detailed description of the ELASPIC method, including cross-validation and comparison with existing approaches can be found in the original publication. A python package implementing the ELASPIC pipeline is hosted on github. Documentation for the ELASPIC pipeline is hosted on readthedocs.

ELASPIC tutorial


Run ELASPIC

On the main input page, users will see the page as below. Follow the numbers to correctly submit a job:

1. Enter protein identifier

Any protein in the Uniprot knowledgebase (>50M) can be entered with the most common identifiers, gene names, and synonyms (>800M). Example input: OO0522, CCM1, KRIT1_HUMAN, ENSP00000391675, GC07M091828.

  • Warning: as only structural models for the human proteome has been precalculated, predictions for other organisms can be expected to last multiple hours per mutation.

2. Select mutation in protein

Mutations can be selected by either clicking the protein figure or by using the select option next to the protein input.

  • Warning: because ELASPIC uses structures of protein domains to predict stability/affinity effects, the mutation has to fall within known domain boundries.

3. Submit job

A valid e-mail address can be entered for notifications about the submitted job. An email will be sent on job start and completion with a link to the results. Check junk e-mail folder in case the notification got delivered there.

! Switch to batch input

Alternative option to input multiple mutations at once or upload a file with data.

!! See previously run mutations

Mutations already in the database are shown on the protein figure. Click each individual mutation to see their results in the same format as on the result page.

ELASPIC prediction results

As output, ELASPIC gives a table of selected mutations and their predicted stability effects, as seen below:

1. Result table

ELASPIC distinguishes between two types of mutations: core (of domain) and interface (between two proteins known to interact). Besides mutation type, the table has six colums with results: (i-ii) The sequence identity and alignment score of the template used to model the protein domain or interaction structure. (iii) The DOPE score of the model given by Modeller. (iv-v) The Gibbs free energy of the optimized wild-type and mutated models calculated by FoldX. (vi) The final change in domain stability (core) or binding affinity (interface) as predicted by ELASPIC.

  • Notice: For interface mutations two values are given for both sequence identity and alignment score. The first one given will always be for the protein entered while the second is for the interacting protein. Furthermore, in the interface multiple interactions might be affected, and multiple rows will then be shown in the result table for the same mutation.

2. Filter results

The result table can be filtered to only show the specific rows desired. Filtering can be done by protein name, status of calculations, type of mutation, and by result scores.

  • Notice: Filtering not only filters the table, but also the downloadable files.

3. Downloadable files

The results can be downloaded in a tab-delimited text file with either only the final predicted energy change or with every single feature used by ELASPIC. Furthermore, created models, alignments, and sequences can be downloaded.

4. See detailed results for mutation

Each completed row can be clicked to see detailed results.

Detailed results

The detailed results include a protein domain graphic and the modelled and wild-type protein structures superimposed in an enhanced dynamic 3D structure viewer:

1. Affected interaction network

All interactions affected by the mutation are included in the network. Each interaction can be clicked to display its detailed results.

  • Notice: For core mutations the interaction network is not shown.

2. Protein interaction figure

The selected interaction is shown as a figure with all protein domains. The interacting domains are color-coded with blue for entered protein and pink for interacting protein.

  • Notice: For core mutations only the entered protein is shown.

3. Modelled structure

A Jmol plugin shows both the modelled wild-type and mutant structures. See mouse controls in their wiki.

! Set quality settings

Jmol runs by default in in Java mode. However, if Java not is installed or compatible with the browser (e.g. mobile phones) the plugin can be switched to HTML5 mode with a single click. Furthermore, if the structural rendering is too demanding for the browser, the quality can be reduced.

!! Set display settings

The buttons below the plugin can manipulate the display. To control details such as labels, colors, and schemes use the buttons on the left. To view or hide a model or protein, or to view the structures in a larger window use the buttons on the right.