DISSPred is a web server that predicts both protein secondary structure and backbone dihedral angles. The theory behind the prediction process was published in P. Kountouris and J.D. Hirst, BMC Bioinformatics 2009, 10:437 and it is highly recommended to read into the background of the methods.

General Tips

DISSPred has simple, easy to use interface. However, to benefit most from its use it is strongly recommended to read through this help section. This is especially important for the options that are used for the predictions. Next to each option on the interface the is a Help small Help link (see to the right) that leads directly to the respective section in this help file.

Job Options

Uploading Files

DISSPred can process FASTA files (plain amino acid sequence). PSI-BLAST will be used first to create the PSSM files. Therefore, it will take time to complete the jobs. It is also possible to upload a number of FASTA files at once by combining them in an archive. Currently the following file types can be handled:

Pasting FASTA sequence

It is possible to copy and paste a single amino acid FASTA sequence. The sequence can only contain the single letter code of the 20 standard amino acids or the letter X for the others. No other characters are accepted.

Job Name

The name of the job is usually made up of the uploaded file name or is given the default file name "sequence" if a FASTA sequence is given. Alternatively it is possible to define a custom job name, which will be turn up in the name of the output directory sent to the user and the main log filename of the job.

Clustering Options

Selecting the clustering algorithm

DISSPred uses two different clustering techniques: Expectation Maximisation (EM) and k-Means. EM was shown to perform slightly better that k-Means. The secondary structure is always predicted using EM clustering with three clusters. Therefore, this field only applies for the prediction of dihedral angles.

Choosing the number of clusters

This field is also used only for the prediction of dihedral angles since the number of clusters is fixed to 3 for secondary structure prediction. The number of clusters for dihedral prediction must be an interger from 2 to 8.

Job Submission

Mail Address

Depending on the number of files submitted, some of the calculations can take from a few minutes to many hours. When FASTA files are submitted, PSI-BLAST must be used to create the PSSM files which is usually time-consuming. Hence, the results are generally only provided by mail.

The output files are all plain text and usually not very large and in general compressed as zip archive. Unless a big number of files is processed, the returned archive is small enough to be sent by mail.

If the compressed folder with the results of the job is bigger than 10 MB (which is very unlikely), a download link is sent via mail instead of the archive itself. If even mails smaller than 10 MB are unwanted or the mail server blocks the DISSPred mails, the sending of a download link can be forced with this tickbox.