For each gene's representative transcript, we considered the [−1000, +1000] bp around the TSS as the putative promoter region, except in the specific analyses listed under Results where an alternate [−1000, 0] bp TSS set was used. We chose relatively large promoter regions with the purpose of incorporating alternative TSS, which in turn allowed us focus on gene-based rather than transcript-based analysis, since alternative promoter usage is a widespread phenomenon in human transcriptome [74], [75]. Although such promoters may incorporate some exonic sequence, it was shown that downstream elements also regulate transcription [76], and therefore including the first kilobase of gene bodies – provided that protein-coding gene properties such as codon bias are controlled for – can provide valuable regulatory information in addition to that residing in the region upstream of the TSS. We obtained the promoter sequences using Galaxy (www.galaxyproject.org/).
