The PSIPRED Protein Sequence Analysis Workbench

The New Beta site is now available. Please consider using the the new site and return any feedback you have to

You can get news and server updates from the new PSIPRED twitter account.

The PSIPRED Protein Sequence Analysis Workbench aggregates several UCL structure prediction methods into one location. Users can submit a protein sequence, perform the predictions of their choice and receive the results of the prediction via e-mail or the web.
For a summary of the available methods you can read More...

NOTE: users who need to run our methods on a large number of proteins should consider downloading our software using the menu on the left (Server Navigation -> Software Download).

Current Contributors David T. Jones, Daniel Buchan, Domenico Cozzetto & Kevin Bryson
Previous Contributors Tim Nugent, Federico Minneci, Anna Lobley, Sean Ward, Liam J. McGuffin

For queries regarding PSIPRED:

Choose Prediction Methods

PSIPRED v3.3 (Predict Secondary Structure) DISOPRED3 (Disorder Prediction)
pGenTHREADER (Profile Based Fold Recognition) MEMSAT3 & MEMSAT-SVM (Membrane Helix Prediction)
BioSerf v2.0 (Automated Homology Modelling) DomPred (Protein Domain Prediction)
FFPred 3 (Eukaryotic Function Prediction) GenTHREADER (Rapid Fold Recognition)
MEMPACK (SVM Prediction of TM Topology and Helix Packing) pDomTHREADER (Fold Domain Recognition)
DomSerf v2.0 (Automated Domain Modelling by Homology)

Input Sequence (Single sequence or Multiple Sequence alignments; as raw sequence or fasta format)

If you wish to test these services follow this link to retrieve a test fasta sequence.

Submission Details

Email Address for job completion alert (optional)


Password (only required for licenced commercial e-mail addresses)


Short identifier for submission



PFilt Options

Select Sequence masking options
Mask low complexity regions
Mask transmembrane helices
Mask coiled-coil regions
Warning: No sequence filters are applied when running MEMSAT or MEMPACK

Homologue Detection

It looks like your input may be a multiple sequence alignment; Multiple alignment input is only available for PSIPRED, pGenTHREADER and pDomTHREADER

Find new close and distant relatives. Your input alignment contains only some of the possible homologues (most accurate calculation)
Find new distant relatives. You are sure your input alignment contains all known close homologues
Find no relatives. You are sure your input alignment contains all known homologues (likely to be least accurate)
Warning: These options only apply to multiple alignment input.

Prediction Options

Include PSIPRED secondary structure prediction
DISOPRED2's False Positive Threshold:
Note: the threshold in False Positive Rate for DISOPRED2 predicitions is now fixed to 5%, which is the default value used by DISOPRED3.

Output Options

Don't return PSI-BLAST output
Return PSI-BLAST hits only
Return PSI-BLAST hits and alignments

Modeller Key:

Input your MODELLER key obtained from the Sali lab


Sequence alignment domain prediction: PSI-BLAST options

Perform PSI-BLAST based prediction: Yes
Select database: PfamA
PSI-BLAST e-value cutoff:
Number of PSI-BLAST iterations:

DomSSEA domain prediction options

Perform DomSSEA prediction: Yes

PSIPRED options

Include secondary structure profile plot: Yes No
Display PSIPRED prediction: Yes No

Ensure you enter your valid MODELLER key via the BioSerf tab.

You input looks like it may be a multiple sequence alignment, please ensure it is in valid fasta format and that all sequences are the same length

You have selected every analysis. The run time will typically be around 5 to 6 hours. You may wish to select fewer specific analyses

Metsite will only analyse one chain in the PDB file at a time

There must be at least 2 chains in your PDB file.

MakeTDB will only analyse single chain PDB files.

Memembed will only process membrane proteins.