Promoter prediction in E. coli based on SIDD profiles and Artificial Neural Networks

Charles Bland,Aleksandra A Markovets,Abigail S Newsome

doi:10.1186/1471-2105-11-s6-s17

Charles Bland, Aleksandra A Markovets + Show 1 more

Open Access

https://doi.org/10.1186/1471-2105-11-s6-s17

Copy DOI

Journal: BMC Bioinformatics	Publication Date: Oct 1, 2010
Citations: 21	License type: CC BY 2.0

Affiliation: Mississippi Valley State University

Abstract

BackgroundOne of the major challenges in biology is the correct identification of promoter regions. Computational methods based on motif searching have been the traditional approach taken. Recent studies have shown that DNA structural properties, such as curvature, stacking energy, and stress-induced duplex destabilization (SIDD) are useful in promoter prediction, as well. In this paper, the currently used SIDD energy threshold method is compared to the proposed artificial neural network (ANN) approach for finding promoters based on SIDD profile data.ResultsWhen compared to the SIDD threshold prediction method, artificial neural networks showed noticeable improvements for precision, recall, and F-score over a range of values. The maximal F-score for the ANN classifier was 62.3 and 56.8 for the threshold-based classifier.ConclusionsArtificial neural networks were used to predict promoters based on SIDD profile data. Results using this technique were an improvement over the previous SIDD threshold approach. Over a wide range of precision-recall values, artificial neural networks were more capable of identifying distinctive characteristics of promoter regions than threshold based methods.

Highlights

One of the major challenges in biology is the correct identification of promoter regions
The current study proposes a more sophisticated approach, involving the use of artificial neural networks (ANNs), along with stress-induced duplex destabilization (SIDD) profiles, for promoter prediction
Training and testing sets construction The training/testing dataset was constructed from the SIDD profile

Summary

Introduction

One of the major challenges in biology is the correct identification of promoter regions. Computational methods based on motif searching have been the traditional approach taken. The currently used SIDD energy threshold method is compared to the proposed artificial neural network (ANN) approach for finding promoters based on SIDD profile data. Wet-lab methods for promoter identification provide accuracy but suffer from being time-consuming. Several computational methods for promoter prediction have been proposed. Most include some analysis of patterns commonly found in promoter regions, such as -10 and -35 motifs [1,2]. These patterns are not always sufficiently conserved to allow for adequate prediction.

Methods

Results

Conclusion