Transcription Start Site Sequencing Strongly Informs Gene Regulatory Network Inference

Understanding of plant gene promoter architecture has long been a difficult challenge due to the lack of large-scale data and analysis methods on the topic.  In this talk I will discuss a large-scale transcription start site (TSS) dataset in Arabidopsis, produced using a high-resolution method for the analysis of 5’ ends of mRNA transcripts.  This technique provides millions of TSS locations from wildtype Col-0 Arabidopsis whole root samples.  Using this quality-filtered dataset, we first categorize the different shapes taken on by the TSS location distributions into TSS “tag clusters”. We then design a model that is able to predict the presence of TSS tag clusters directly from genomic sequence with high accuracy for each cluster shape.  We use this model to analyze the transcription factor binding site content of the promoter sequences for each TSS tag cluster shape category.  In contrast to the canonical notions of sharp narrow peak TATA-containing promoters vs more broad “TATA-less” promoters, the model demonstrates that a large compendium of known DNA sequence binding elements is actually necessary and sufficient for accurate promoter prediction in the case of all tag cluster shapes.  These elements form ‘promoter signatures’ for transcription initiation. By identifying the specific combinations and relative locations of known binding elements that lead to transcription, the promoter signatures allow for dramatically improved prediction of transcriptional network interactions in plants where only genomic sequence information is available.