Development and validation of a new PCR optimization method by combining experimental design and artificial neural network.

Polymerase chain reaction (PCR) is one of the most powerful techniques in a variety of clinical and biological research fields. In this paper, a chemometrics approach, combining experimental design (ED) and artificial neural network (ANN), was proposed for optimization of PCR amplification of lycopene cyclase gene carRA in Blakeslea Trispora. Five-level star design was carried out to obtain experimental information and provide data source for ANN modeling. Nine variables were used as inputs in ANN, including the added amount of template, primer, dNTP, polymerase and magnesium ion, the temperature of denaturating, annealing and extension, and the number of cycles. The output variable was the efficiency (yield) of the PCR. Based on the developed model, the effects of each parameter on PCR efficiency were predicted and the most suitable operation condition for present system was determined. At last, the validation experiment was performed under the optimized condition, and the expectant results were produced. The results obtained in this paper showed that the combination of ANN and ED provided a satisfactory optimization model with good descriptive and predictive abilities, indicating that the method of combining ANN and ED can be a useful tool in PCR optimization and other biological applications.