ACCURATE PREDICTION OF PROTEIN SECONDARY STRUCTURAL CLASS WITH FUZZY STRUCTURAL VECTORS

The prerequisites for accurate prediction of protein secondary structu ral class (all-alpha, all-beta, alpha+beta, alpha/beta or multidomain) were studied, and a new similarity-based method is presented for the prediction of the secondary structural class of a protein from its seq uence. The new method uses representatives of nuclear families as a le arning set. For the sequence to be predicted, the method produces a ve ctor of certainty factors called a fuzzy structural vector, Validation with independent test sets shows that the prediction accuracy of the proposed method has clear dependency on the representativity of the le arning set. The representatives obtained from the nuclear families of the Brookhaven Protein Data Bank (PDB) were shown to give accurate pre dictions for PDB proteins, whilst the amino acid composition-based met hods used previously achieve their maximum predictability with relativ ely limited learning sets, and they remain inaccurate even with highly representative learning sets. The usability of the new method is incr eased further by the fuzzy structural vectors, which substantially red uce the risk of misclassification and realistically describe vague sec ondary structural tendencies.