Title
Bi-billboard: symmetrization and careful choice of informant species results in higher accuracy of regulatory element prediction.
Abstract
The identification of cis-regulatory modules (CRM) is one of the most important problems towards the understanding of transcriptional regulation in higher eukaryotes. Computational methods for CRM detection are gaining importance due to the availability of genomic data on one side, and costs and difficulties of experimental methods on the other side. One of proposed approaches, called Billboard, predicts CRMs based on the location of transcription factor binding sites in an analyzed sequence and a related one in so-called informant species. In the present article, we show how to combine information obtained in two symmetric runs (on the sequence of interest and on the related one) of the Billboard tool. In a series of experiments on data from various organisms, we show that the predictive power of our symmetric approach is significantly higher than the power of the one-way approach of Billboard. Moreover, we show that the evolutionary distance between organisms considerably influences the quality of prediction and we provide guidelines on the choice of an informant species.
Year
DOI
Venue
2011
10.1089/cmb.2010.0299
JOURNAL OF COMPUTATIONAL BIOLOGY
Keywords
Field
DocType
computational molecular biology
Predictive power,Computational molecular biology,DNA binding site,Symmetrization,Artificial intelligence,Bioinformatics,Score,Machine learning,Mathematics
Journal
Volume
Issue
ISSN
18.0
6
1066-5277
Citations 
PageRank 
References 
0
0.34
10
Authors
3
Name
Order
Citations
PageRank
Norbert Dojer11599.44
Przemyslaw Biecek235.46
Jerzy Tiuryn31210126.00