Abstract—Type II membrane proteins in the Golgi apparatus play important roles in biological functions, and predominantly exist as catalysts related to post-translational sugar modification. This study describes a new method for detecting Golgi-localized type II membrane proteins (GLs) from post-Golgi type II membrane proteins (PGs), which are mainly localized in the plasma membrane and endoplasmic reticulum (ER). The method is based on hydropathy profiles and the position-specific scoring matrix (PSSM) in combination with the back propagation artificial neural network (BP-ANN). The accuracy of discriminating GLs from PGs was evaluated in a 5-fold cross-validation test with 94.7% sensitivity and 93.5% specificity. This result shows that our method can predict GLs with high accuracy, and that the PSSM and BP-ANN combination can effectively discriminate GLs.

Index Terms—Golgi, endoplasmic reticulum (ER), plasma membrane, type II membrane protein, discrimination, hydropathy analysis, position-specific scoring matrix (PSSM)

Cite:Tatsuki Kikegawa, Kenji Etchuya and Yuri Mukai, "Neural Network-Based Discrimination of Golgi Type II Membrane Proteins with Better Accuracy," International Journal of Pharma Medicine and Biological Sciences, Vol. 7, No. 1, pp. 16-19, January 2018. doi: 10.18178/ijpmbs.7.1.16-19