Schematic outline of the data analysis strategy. (a) Extraction of promoter sequences and generation of random data sets. For each annotated protein or peptide-coding gene the translation initiation codon (ATG) was located and a 1-kb upstream sequence was extracted. The genomic data set thus obtained contained 26,140 entries for Arabidopsis thaliana and 49,472 entries for Oryza sativa. For each species 100 independent randomised data sets were generated by reshuffling the nucleotide sequence of each entry. (b) In-silico screening using a CRE-specific frequency matrix. A positive hit was recorded if the matrix score reached a 95% cut-off threshold. (c) Determination of the distance of an element to the ATG. In the illustrated example d1 and d3 are the distances to the ATG for the elements localized on the [+] strand; d2 is the distance to the ATG for the element localized on the [-] strand. (d) Gap length between elements. In the example above Δ1 is the gap between two cis-oriented CREs (in this case [+/+]), and Δ2 and Δ3 are gaps between two trans-oriented CREs [-/+] and [+/-], respectively.