Adding sequence context to a Markov background model improves the identification of regulatory elements

• DOI: 10.1093/bioinformatics/btl528
• Publication Date: 2006-10-27T00:26:52Z
• Authors: Nak-Kyeong Kim; Kannan Tharakaraman; John L. Spouge
• Keywords: Identification; Maximum-entropy Markov model; Sequence (biology)

Abstract Motivation: Many computational methods for identifying regulatory elements use a likelihood ratio between motif and background models. Often, the model of independent bases. At least two different Markov models have been proposed with aim increasing accuracy predicting elements. Both suffer theoretical drawbacks, so this article develops third, context-dependent from fundamental statistical principles. Results: Datasets containing known in eukaryotes provided basis comparing predictive accuracies Non-parametric tests indicated that order 3 constituted statistically significant improvement over Our performed slightly better than previous We also found discriminating competing models, correlation coefficient is more sensitive measure performance coefficient. Availability: C++ program available at Contact: [e-mail] Supplementary information: data are Bioinformatics online.