Application of expert networks for predicting proteins secondary structure

The presentstudy utilizes expertneural networksforthe prediction of proteinssecondary structure. Weuse threeindependent networks, one for each structure (alpha, beta and coil) asthe first-level processing unit;decision upon the chosenstructure for each residue is carried out by a secondlevel, post-processingunit,whichutilizes the Chouand Fasman frequency values FaandFb inorder tostrengthen and/ordeplete theprobabilityof the specific structure under investigation. The highest prediction case was 76%. Our method requires primitive computational means and a relatively small training set, while still been comparable to previous work. It is not meant to be an alternative to the determination of secondary structure by means of free energy minimization, integration of dynamic equations of motionorcrystallography,which areexpensive, time-consumingandcomplicated,buttoprovideadditionalconstrains,whichmightbeconsidered and incorporated into larger computing setups in order to reduce the initial search space for the above methods. # 2006 Elsevier B.V. All rights reserved.