Deriving the Upstream Truncation Thresholds of Human Promoter Sequences Based on Expression Correlation

Most promoter sequences are generated by arbitrary truncating regions upstream from the transcription start site, typically around 1,000 base pairs. However, this arbitrary approach might miss essential regulatory elements required for proper transcription. Unfortunately, there is no widely accepted rationale for choosing a specific promoter truncation threshold, despite its importance in obtaining expected expression profiles. In this study, we present a data-driven method to define biologically meaningful promoter thresholds. We begin by identifying putative shared promoters between neighboring gene pairs and classify them as true shared promoters based on expression similarity using GTEx data. Our analysis reveals that shorter putative shared promoters are more likely to be true shared promoters and thus lead to similar expression profiles. Using the lengths of these true shared promoters, we subsequently define aggressive, standard, and conservative thresholds for human promoter sequences. We further refine these thresholds using known cis-regulatory elements and transcription factor binding regions, resulting in more compact promoter boundaries. These refined thresholds provide a biologically grounded alternative to arbitrary values and establish a foundation for future experimental validation and practical applications. Moreover, our refined thresholds are especially well-suited for scenarios where compact promoter design is critical.