DSC: Discrimination of Protein Secondary Structure Class
Ross D. King & Michael J.E. Sternberg
NEW Version 1.2
DSC now includes an updated and more accurate GOR matrix. This has
been found to improve prediction accuracy by 0.5-1%. The matrix is
derived from a dataset of 496 proteins compared to the original
dataset of 126. This new dataset is based on one developed by Geoff
Barton (EBI) and is almost a proper superset of the original 126. The
reason it is not quite a superset is that the definition of homology
used is now stricter). The dataset differs from that of Barton in
excluding proteins with less than 30 residues.
In addition the homology database has been greatly enlarged and
updated, DSC now uses Blast 2 to carry-out the homology search.
We estimate that these changes make a 1%-2% improvement in prediction
accuracy.
The upgrade to DSC was funded by BBSRC/EPSRC Bioinformatics initiative
grant no. BIF08765.
