Authors: David T. Jones
Source: [Protein secondary structure prediction based on position-specific scoring matrices]
PSIPRED is a secondary structure prediction tool, which uses neural networks. The neural network has a single hidden layer and a feed-forward back-propagation architecture. The procedure of this method is split into three main steps. The first one is the generation of sequence profiles which means it generates a position-specific scoring matrix from PSI-BLAST and takes it as an input for the neural network. The second step is the prediction of the initial secondary structure which means it creates an output layer where the units represent one of three secondary structure states (helix, strand or coil). The last step is the filtering of the predicted structure which is the successive filtering of the outputs from the main network.
We used the [Webserver] for our analysis. The input for the webserver is only the sequence in FASTA-format.
As a prediction result you get different possible files with the predicted secondary structure. The possibles outputs are in pdf, postscript or txt. The pdf and postscript have more graphical representation of the result whereas the txt is simpler.
Authors: Cole C, Barber JD, Barton GJ
Source: [The Jpred 3 secondary structure prediction server]
Jpred3 is a server for secondary structure prediction. It uses the Jnet algorithm for the prediction which consists of serveral neural networks. The special thing about is that it has two possible inputs: the sequence or a multiple sequence alignment. Furthermore it delivers different possible output files like HTML, pdf or postscript.
We used the [Webserver] for our analysis. The input for the webserver was in our case only the sequence in FASTA-format.
As a prediction result we got different possible files with the predicted secondary structure. There are some complex and some simple outputs. The complex ones contains the multiple sequence alignment as well as the predicted secondary structure. In contrast the simple ones contain only the secondary structure prediction.
Authors: Kabsch W, Sander C.
Source: [Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features.]
DSSP is a database for secondary structure assignments for each PDB entry. It is no prediction tool, but is often used to determine the prediction success by comparing the predicted secondary structure with the one from DSSP. It defines the secondary structure by given atomic coordinates in PDB-format. It bases mainly on H-bonding, because there are specific h-bonds at more or less specific positions which define helices or sheets.
We used the [Webserver] for our analysis. There are two possible inputs: the PDB-id or the sequence. We used the PDB-id.
As a result you get a file with the assigned secondary structure, the symmetry and the accessibility of the whole protein.