Changes in v/1.1
The training dataset has been slightly enlarged to include more
'negative' (non-glycosylated) sites.
Surface-derived threshold has been modified to perform a running
Glycosylation potentials have been sigmoidally scaled to magnify the
distinction between positive and negative sites.
We expect the server to predict sites more 'specifically', that is to pick
upO-GlcNAc sites fromsecreted and membrane proteins of
D.discoideum. Predictions with the new server may vary from those made
with DictyOGlyc v/1.0. On a general basis, the server picks fewer glycosylation
sites presumably excluding more false positives.
Feedback, Comments and suggestions are most welcome.