A neural computing method for identifying quantitative structure activity relationships

A novel genetic algorithm for neural computing to identify quantitative str ucture activity relationship (QSAR), named mutation - based genetic algorit hm (MGA), is presented. MGA only uses the mutation operator for local searc h. To enhance the efficiency of local search, the genes that represent the variables employ different time - varying mutation rates in MGA. Combining random restart technique with the local search strategy, the algorithm can give satisfactory solution in a limited time. As a typical object of the neural computing for QSAR, a set of 74 2,4 - dia lmino - 5 - (substituted benzyl) pyrimidines that inhibit dihydrofolate red uctase were used to verify the effectiveness of MGA in computings of predic ting bio - activity. Cross - validation trials and the test of predicting a ctivity demonstrated that the predictive ability of the QSAR model built wi th MGA is better than those provided by other methods.