Transmembrane helices topology prediction: using a simplified transmembrane HMM

• Published in: International Conference on Neural Networks and Signal Processing, 2003. Proceedings of the 2003 Vol. 1; pp. 571 - 574 Vol.1
• Main Authors: Ming-hui Wang, Ao Li, Tao Wang, Yun Zhou, Huan-qing Feng
• Format: Conference Proceeding
• Language: English
• Published: IEEE 2003
• Subjects: Biological system modeling; Biological systems; Biomedical engineering; Biomembranes; Computational efficiency; Hidden Markov models; Predictive models; Protein engineering; Tail; Topology
• ISBN: 9780780377028; 0780377028
• DOI: 10.1109/ICNNSP.2003.1279337

The hidden Markov model (HMM) based on proper architecture corresponding to the biological systems is presented to model and predict the location and orientation of alpha helices in membrane transmembrane proteins. The StHMM (segment trained HMM) is composed of five sub-HMMs with their own independent structures corresponding respectively to helix core, loop on the cytoplasmic side, short and long loops on the non-cytoplasmic side, and globular on each side. Since the standard BM algorithm is a local optimizing progress and exhaustive searching way, it can be improved by taking advantage of the property of the transmembrane with location information. Using the new method, we got 86.88% accuracy of the entire correct location (without orientation) topologies in a dataset of 160 proteins with known topology. Computation cost is reduced in the meantime.