The ReFOLD server help page
This page contains simple guidelines for using the ReFOLD server, sample input data which may be downloaded and submitted and examples of output from the server.Guidelines for using the server
The ReFOLD server requires the amino acid sequence of your target protein and a single 3D model file in PDB format.If you decide to provide your email address then you will be sent a link to the graphical results and machine readable results when your refinement has completed. However, if you do not provide your email address then you must bookmark the results page and so that you may refer to the results when they become available. The job status for your target will be reported on the results page prior to job completion.
Required - Sequence Data
In the text box labelled "Input sequence of protein target" please carefully paste in the full amino acid sequence for your target protein in single letter format. An example sequence is shown below:
Sample sequence (CASP target TR594):
KINQPEHLAQLDGYSQKKGISGAHNADVFNKAVVDNGVK IISETPTGVRGITQVQYEIPTKDAAGNTTGNYKGNGAKP FEKTIYDPKIFIt is important that you provide the full sequence that corresponds to the sequence of residue coordinates in the model file. If your model does not contain numbering that corresponds directly to the order of residues in the sequence file then the server will attempt to renumber the residues in the model files accordingly. However, if there are residues in a model file that are not contained in the provided sequence then the prediction for that model will not complete.
Required - Model Data
Using the file selector labelled "Upload model/models" you may only upload a single PDB file. Please ensure that each PDB file contains the coordinates for one model only.
The server will attempt to automatically renumber the ATOM records in each model in order to match the residue positions in the sequence i.e. the coordinates for the first residue in the sequence will be renumbered "1" in each model file (if they aren't already), the coordinates for the second residue in the sequence will be numbered "2", and so on.
Sample PDB file:
An example file containing a single model for the sample sequence mentioned above can be downloaded below:
Example of a starting model - CASP12 refinement target TR594: REFINEMENT TARGET TR594
Optional - E-mail address
If you wish, you may provide your e-mail address. You will be sent a link to the graphical results and machine readable results when your predictions are completed.
Privacy Notice: Processing of personal data will be in accordance with the GDPR and University of Reading (UoR) Data Protection Policy. Users' IP addresses will be temporarily stored in the queuing system and then used to generate anonymous usage statistics. Optionally, users may provide an email address, so that they can be notified when their job completes; this will be deleted when no longer required. Personal data will be accessible only by UoR staff managing the server, and will not be stored for longer than is necessary for the provision of the service.
Optional - Short name for sequence
If you wish, you may assign a short memorable name to your prediction job. This is useful so that you can identify particular jobs in your mailbox. This is particularly important because ReFOLD will not necessarily return your results in the order you submitted them. The set of characters you can use for the filename are restricted to letters A-Z (either case), the numbers 0-9 and the following other characters: .~_- The name you specify will be included in the subject line of the e-mail messages sent to you from the server.
Output from the server
The ReFOLD server version 3.0 produces graphical output for the top selected refined model, followed by table containing numerical and graphical prediction results for all of the refined models generated by ReFOLD.
Examples of graphical output:
Full ReFOLD results for CASP target TR594
Full ReFOLD results for CASP target R1055
The top ranked model selected by ReFOLD protocol is shown in the first table together with the starting model and error per-residue plot. The download links to the machine-readable output generated by ReFOLD are also provided in CASP format
Examples of machine-readable outputs:
Quality assessement data file for refined models for CASP target TR594
Compressed archive of refined models generated by ReFOLD for CASP target TR594
The results table is ranked according to decreasing global model quality score. The global model quality scores range between 0 and 1. In general, scores less than 0.2 indicate there may be incorrectly modelled domains and scores greater than 0.4 generally indicate more complete and confident models, which are highly similar to the native structure. If the global model quality scores are low, then the per-residue scores can give you an idea of specific domains or regions in your protein that might be correctly modelled.
From the global scores we can calculate a p-value which represents the probability that each model is incorrect. That is to say, that for a given predicted model quality score, the p-value is the proportion of models with that score that do not share any similarity with the the native structure (TM-score < 0.2). Each model is also assigned a colour coded confidence level depending on the p-value:
| P-value cut-off | Confidence | Description |
| p < 0.001 | CERT | Less than a 1/1000 chance that the model is incorrect. |
| p < 0.01 | HIGH | Less than a 1/100 chance that the model is incorrect. |
| p < 0.05 | MEDIUM | Less than a 1/20 chance that the model is incorrect. |
| p < 0.1 | LOW | Less than a 1/10 chance that the model is incorrect. |
| p > 0.1 | POOR | Likely to be a poor model with little or no similarity to the native structure. |
| Improvement in global score (%) | Color | Description |
| ReFOLD model > starting model | Positive value | Indicates that the global score has been improved. |
| ReFOLD model < starting model | Negative value | Indicates that the global score has not been improved. |
The per-residue scores indicate the predicted distance (in Angstroms) between the CA atom of the residue in the model and the CA atom of the equivalent residue in the native structure. Thumbnail images of plots depicting the per-residue error versus residue number are included in each row in the results table. Each of the thumbnails links to a page that displays a larger view of the plot and contains a further link to download a PostScript version. Each row in the table also displays a thumbnail of the 3D cartoon view of the model which is colour coded with the residue error according to the RasMol temperature colouring scheme. Each small image also links to a page that shows a larger image of the 3D view and contains a link to download a PDB file of the model with residue accuracy predictions (Angstroms) in the B-factor column. The orignal and refined models are also loaded into JSmol for convenient interactive viewing of per-residue errors within the browser.
How long will I have to wait for my results?
The time taken for ReFOLD to complete its protocol will depend on the size of the submitted model and the load on the server. For example, for a starting model of ~100 residues, you should receive your results back within ~48 hours once the job is running.
Fair usage policy
You are only allowed to have 1 job running at a time for each IP address, so please wait until your previous job completes before submitting further data. If you already have a job running then you will be notified and your uploaded data will be deleted. Once your job has completed your IP address will be unlocked and you will be able to submit new data.
Error reporting
Check the header of the machine readable results file (provided as a link at the top of the result page) for any errors that may have occured following file submission. Please email me for help if you encounter a persitent error.