A simple and fast secondary structure prediction method using hidden neural networks

• DOI: 10.1093/bioinformatics/bth487
• Publication Date: 2004-09-18T00:13:19Z
• Authors: Kuang Lin; Victor A. Simossis; Willam R. Taylor; Jaap Heringa
• Keywords: Sequence (biology)

Abstract Motivation: In this paper, we present a secondary structure prediction method YASPIN that unlike the current state-of-the-art methods utilizes single neural network for predicting elements in 7-state local scheme and then optimizes output using hidden Markov model, which results providing more information prediction. Results: was compared with top-performing methods, such as PHDpsi, PROFsec, SSPro2, JNET PSIPRED. The overall accuracy on independent EVA5 sequence set is comparable of top performers, according to Q3, SOV Matthew's correlations measures. shows highest terms Q3 scores strand Availability: available on-line at Centre Integrative Bioinformatics website (http://ibivu.cs.vu.nl/programs/yaspinwww/) Vrije University Amsterdam will soon be mirrored Mathematical Biology (http://www.mathbio.nimr.mrc.ac.uk) NIMR London. Contact: [e-mail]