Independent component analysis-based prediction of O-Linked glycosylation sites in protein using multi-layered neural networks
- Resource Type
- Conference
- Authors
- Wang, Chu-Zheng; Tan, Xiao-Feng; Chen, Yen-Wei; Han, Xian-Hua; Ito, Masahiro; Nishikawa, Ikuko
- Source
- IEEE 10th INTERNATIONAL CONFERENCE ON SIGNAL PROCESSING PROCEEDINGS Signal Processing (ICSP), 2010 IEEE 10th International Conference on. :1-4 Oct, 2010
- Subject
- Signal Processing and Analysis
Components, Circuits, Devices and Systems
Communication, Networking and Broadcast Technologies
Computing and Processing
Artificial neural networks
Protein sequence
Principal component analysis
Accuracy
Pattern analysis
Encoding
O-glycosylation
pattern analysis
positional probability function
independent component analysis
multi-layer neural network
- Language
- ISSN
- 2164-5221
2164-523X
In this paper, we develop a new method for prediction 0-linked glycosylation site and pattern analysis in protein, which combines independent component analysis (ICA) with a multi-layer neural network (NN). ICA is first used to construct main basis (subspace) of the protein sequence for features extraction. The projections of protein sequence on the subspace with low dimension are used as input data instead of the higher-dimensional protein sequences. Neural network is built to predict whether a particular site of serine or threonine is glycosylated. Compared with other subspace method, our proposed new method can improve the prediction accuracy.