We have systematically analyzed a set of protein disordercausing
mutations and observed that 88% of mutants destabilize
the proteins. A set of 208 features were derived using the information
on wild-type residue, mutant residue, adjacent residues on
both sides of the mutant, amino acid properties, mutation matrices,
and statistical potentials. The feature selection method was
employed to select the best features and develop an SVM-based
model using the selected 9 features to discriminate the mutants
into O / D or O / O class. Our method achieved a classification
accuracy of 90.0% with sensitivity and specificity of 94.9 and 80.6%,
respectively, on a dataset of 90 mutants. The performance of the
current method is superior to that of other methods in the literature.
Hence, we suggest that our method (DIM-Pred) could be used
for predicting the disorder-related mutants with high efficiency.