Computational analysis of promoter features associated with contrasting expression patterns of human genes under normal and disease conditions

The promoter region plays a critical role in regulating gene expression, particularly in complex vertebrate genomes. However, promoters associated with contrasting gene expression patterns have not been comprehensively characterized. In this study, we systematically analyzed multiple features of human promoter sets corresponding to contrastingly expressed gene sets or transcripts (PCeGs). The promoters considered were those of genes transcribed only in specific tissues (brain, spleen, and testis), as well as those transcribed ubiquitously under normal conditions, and those up-regulated, down-regulated, or undifferentiated in specific disease conditions. Four cancers (breast, lung, prostate, and acute myeloid leukemia) and four non-cancer conditions (Crohn's disease, non-obstructive azoospermia (NOA), schizophrenia, and tetralogy of Fallot) were considered. We observed consistent differences in the values of many promoter features across several pairs of PCeGs studied. These features included lower GC/CG content and BREu prevalence, as well as fewer CpG islands in down-regulated (compared to up-regulated) and spleen-specific (compared to ubiquitous or other tissues) PCeGs. Many PCeGs also exhibited a differential abundance of specific transcription factor binding sites (TFBSs). Furthermore, seven DNA structural properties (free energy, electrostatic potential, opening, propeller twist, stagger, slide, and stretch) exhibited significant differences between the PCeG sets. Our findings could help to develop more reliable means for computational characterization, differentiation, and even prediction of promoters associated with differently expressed genes.