4. Conclusions 
We have outlined our strategy of “upstream analysis,” which is an integrated promoter and pathway analysis. The largest part of this analysis has been put together as a workflow in the geneXplain platform. Part of its efficiency is due to a novel approach to identify enriched transcription factor binding sites, which improves the ranking of true motifs according to the (corrected) Yes/No ratio, specifically for suboptimal motif patterns as validated on a large number of ChIP-seq datasets. Here we present two formulas to calculate the correction which provide substantial speed improvements over our previous method. We have compared different methods to obtain the best ranking of motifs and found that Yes/No ratio correction improves the ranking of true motifs, where the confidence interval-based correction is simple to compute and performed comparably to a method making use of the Beta distribution. When we applied our strategy to clustered gene sets of liver or lung tissue that were exposed to a toxicant (naphthalene), we were able to identify tissue-specific targets and master regulators. In the case of liver, these master regulators indicate that some general tumor and apoptosis-promoting pathways may be triggered, whereas in the lung tissue, master regulators were found that specifically trigger aggressive lung cancer to develop. These results demonstrate the validity of the presented upstream analysis strategy.