Positive and Negative Sequence Data Across Organisms

We compiled promoter sequence datasets for eight organisms, each containing experimentally validated positive promoter sequences. For each organism, two distinct negative datasets were generated: one by shuffling the positive promoter sequences while preserving nucleotide composition (positive-promoter-shuffled), and another by randomly sampling non-promoter genome fragments (random–non–promoter–fragments). The corresponding files are named using the suffixes substitution and genome, respectively, where substitution indicates the shuffled method and genome indicates the random sampling method.