Resilient Back Propagation based Yield Prediction of Keratinase from Bacillus Megaterium SN1

K. Asawa, N. Wadhwa, and S. Agrahari (India)

Keywords

Keratinase, Bacillus megaterium SN1, ResilientBackpropagation (RPROP)

Abstract

A Bacillus strain was isolated from soil of a regular feather dumping site of Ghazipur poultry processing plant, Ghaziabad, India. Strain was identified as Bacillus megaterium SN1 after morphological, biochemical and cultural characteristic and was found to produce keratinase extracellularly in the media. Cell free media of twelve experimental setups that varied in their medium components viz. NaCl, Yeast extract and Feather were checked for specific activity of keratinase. To obtain the best prediction the several variants of feed forward by the neural network architectures were trained and evaluated using these obtained 12 experimental data-set. The trained network could predict specific activity of 15 other experimental setups and we report that Resilient Backpropagation (RPROP) was the best architecture with the order of error 1.88e-18. Codes were developed by MATLAB 2007. Optimum production of Keratinase forecasted according to RPROP studies are NaCl - 0.5g, Yeast - 0.1g and Feather – 10 g. These combinations yielded a specific activity of 17.23 units/mg protein.

Important Links:



Go Back