Choosing the Optimal Hidden Markov Model for Secondary-Structure Prediction

Authors:
Juliette Martin;Jean-Francois Gibrat;Francois Rodolphe
Affiliations:
French National Institute of Agriculture Research;French National Institute of Agriculture Research;French National Institute of Agriculture Research
Venue:
IEEE Intelligent Systems
Year:
2005

Citing 0
Cited 3

Combining segmental semi-Markov models with neural networks for protein secondary structure prediction

Neurocomputing
Adaptive Exact Inference in Graphical Models

The Journal of Machine Learning Research
PSSP with dynamic weighted kernel fusion based on SVM-PHGS

Knowledge-Based Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

Researchers have developed many methods to predict proteins' secondary structure solely on the basis of their sequences. Most of these methods rely on neural networks, which offer good accuracy but are hard to interpret. An alternative method aims to find an optimal hidden Markov model to classify protein residues into secondary-structure classes. In addition to producing models that are more easily interpreted, HMMs provide a probabilistic framework for sequence treatment. The model developed with this method features 36 hidden states and offers a compromise between prediction accuracy and a reasonable number of parameters.This article is part of a special issue on data mining for bioinformatics.