The RNA polymerase II (Pol II) core promoter is the strategic site of convergence of the signals that lead to the initiation of DNA transcription^1-5, but the downstream core promoter in humans has been difficult to understand^1-3. Here we analyse the human Pol II core promoter and use machine learning to generate predictive models for the downstream core promoter region (DPR) and the TATA box. We developed a method termed HARPE (high-throughput analysis of randomized promoter elements) to create hundreds of thousands of DPR (or TATA box) variants, each with known transcriptional strength. We then analysed the HARPE data by support vector regression (SVR) to provide comprehensive models for the sequence motifs, and found that the SVR-based approach is more effective than a consensus-based method for predicting transcriptional activity. These results show that the DPR is a functionally important core promoter element that is widely used in human promoters. Notably, there appears to be a duality between the DPR and the TATA box, as many promoters contain one or the other element. More broadly, these findings show that functional DNA motifs can be identified by machine learning analysis of a comprehensive set of sequence variants.

DNA sequence signals in the core promoter, such as the initiator (Inr), direct transcription initiation by RNA polymerase II. Here we show that the human Inr has the consensus of BBCA₊₁BW at focused promoters in which transcription initiates at a single site or a narrow cluster of sites. The analysis of 7678 focused transcription start sites revealed 40% with a perfect match to the Inr and 16% with a single mismatch outside of the CA₊₁ core. TATA-like sequences are underrepresented in Inr promoters. This consensus is a key component of the DNA sequence rules that specify transcription initiation in humans.