Prediction of GFP spectral properties using artificial neural network

Abstract: In this study, we applied artificial neural network, implementing the backpropagation algorithm, for the prediction of the excitation and emission maxima of green fluorescent protein as a function of the quantum chemical descriptors. The data set comprising of 2 sets were collected from the literature. The first data set consisted of GFP color variants, while the second were composed of synthetic chromophores. The modeling of excitation and emission maxima gave good accuracy as observed from the correlation coefficient of 0.9. Results indicated that among different types of molecular descriptors, quantum chemical descriptors yielded the best correlation with experimental data. Furthermore, the protonation state of GFP chromophores were shown to be crucial toward the predictive performance. A comparative analysis with other multivariate methods indicated that artificial neural network was suitable for modeling the spectral properties of the chromophores. Furthermore, this appr...

• Type: Journal
• Year: 2007
• Where: JCC
• Authors: Chanin Nantasenamat, Chartchalerm Isarankura-Na-Ayudhya, Natta Tansila, Thanakorn Naenna, Virapong Prachayasittikul
• Keywords: Artificial Neural Network, Chromophores, Emission Maxima, JCC 2007