ABSTRACT:
This study is focused on optimization of phytase production by Hypocrea lixii SURT01, a fungal isolated from soil. The changes in the concentration of carbon source (sucrose), nitrogen source (peptone), substrate (phytate) and pH using Response Surface Methodology - Box-Behnken (RSM-BB) showed significant changes in the production of phytase. 1.0% sucrose, 0.2% phytate, 3.0% peptone and pH 5.5 were found to be optimum levels of variables and this combination resulted in 87.77 U of phytase production after 96 h of incubation while the predicted output was 83.34 U/ml. The optimized medium resulted in significant increase of the phytase yield by Hypocrea lixii SURT01 in shake flask cultivation. Artificial Neural Networks (ANN) was used to assess the output given by the above model. The effect of four different variables in different concentration on enzyme production was studied. The data received from RSM-BB analysis were used to educate ANN and 20 responses were used to test the trained network. This network was designed in a backward neural network with four neurons in the input layer and one neural in output layer and optimum concentration of variables for phytase was identified using the trained network. The regression coe?cient (R2) showed a good correspondence between predicted and actual data sets for both trained (93.35) and validated data (0.965). A multilayer feed forward ANN educated by back propagation algorithms was designed to predict the productivity of phytase enzyme. The final optimal ANN model obtained using RSM was lucratively learned the association between the input parameters and outputs parameter. The association between actual and predicted result was another proof of the high performance of ANN for estimation of enzyme-catalyzed reaction and hence this process can be applied in estimating the release of phosphate from phytate-bound soil using Hypocrea lixii SURT01.