comprehensive human transcription factor binding site map for combinatory binding motifs discovery全面的人类转录因子结合位点组合绑定主题地图的发现.pdfVIP

Comprehensive Human Transcription Factor Binding Site Map for Combinatory Binding Motifs Discovery ¨ 1 ¨ 2,3 ´ 1 Arnoldo J. Muller-Molina , Hans R. Scholer , Marcos J. Arauzo-Bravo * ¨ 1 Computational Biology and Bioinformatics Group, Max Planck Institute for Molecular Biomedicine, Munster, Germany, 2 Department of Cell and Developmental Biology, ¨ ¨ ¨ Max Planck Institute for Molecular Biomedicine, Munster, Germany, 3 Medical Faculty, University of Munster, Munster, Germany Abstract To know the map between transcription factors (TFs) and their binding sites is essential to reverse engineer the regulation process. Only about 10%–20% of the transcription factor binding motifs (TFBMs) have been reported. This lack of data hinders understanding gene regulation. To address this drawback, we propose a computational method that exploits never used TF properties to discover the missing TFBMs and their sites in all human gene promoters. The method starts by predicting a dictionary of regulatory ‘‘DNA words.’’ From this dictionary, it distills 4098 novel predictions. To disclose the crosstalk between motifs, an additional algorithm extracts TF combinatorial binding patterns creating a collection of TF regulatory syntactic rules. Using these rules, we narrowed down a list of 504 novel motifs that appear frequently in syntax patterns. We tested the predictions against 509 known motifs confirming that our system can reliably predict ab initio motifs with an accuracy of 81%—far higher tha