From Bioinformatikpedia


Basic Information

Author Ward JJ, Sodhi JS, McGuffin LJ, Buxton BF, Jones DT
Year 2004
Reference PubMed 15019783
ML Method SVM

DISOPRED generates a PSI-BLAST profile of the input sequence. A SVM is feeded with a window of 15 residues for each residue of the profile. This SVM predicts if the residue in the center of the window is in an ordered or disordered region of the protein.

The SVM was trained on 750 non-redundant X-ray structures with high resolution. It was assumed that residues which appear in the sequence but whose coordinates from the electron density map are missing are in a disordered region of the protein. For each structure was a PSI-Blast profile calculated. These profiles were used in windows of 15 residues to train a SVM. This way of defining disordered regions in a protein has its disadvantages, but there are no high-throuput methods to measure disordered regions.


How to run it locally?

Change in the disopred script the following lines:

# Change database location

set dbane = /data/blast/nr/nr

# Check if there is any path with /usr/local/bin

change all path to /usr/bin (important: change also the path to the perl location)