Unraveling Regulatory Interactions by the Integration of Genome-wide Location Data and Mutant Data

The ability to model complex regulatory networks is a big challenge in current biology research. Recently many computational methods have been developed to address this challenge. Compared with those methods using single data source, approaches integrating multiple data sources are expected to identify more reliable regulatory relations. Here, we present an easy but powerful method to integrate genome-wide location data and TF (Transcription Factor) mutant data. The integration of these two data, which investigate complementary (physical and functional) aspects of transcription respectively, provides strong evidences of relations between TF and their targets. However, quite low overlap of these two data has hindered their combination. To improve the overlap and simultaneously find reliable interactions, we try to seek the optimal combination of these two data based on hypergeometric distribution. We demonstrate our method on yeast data and validate our predictions by YEASTRACT, high quality ChIP-chip data and other literatures. The results show our method is applied successfully on identifying regulatory interactions. Even with low quality ChIP-chip data, our method uncovers more relations with less false positives than those with high quality data. Furthermore, our method also shows a good performance on discovering cooperative TF pairs.