Finding functional promoter motifs by computational methods: a word of caution

The standard practice in the analysis of promoters is to select promoter regions of convenient length. This may lead to false results when searching for Transcription Factor Binding Sites (TFBSs), since the sequences may contain coding segments. In such cases, motif detection may single out motifs from the coding regions. The mapping of TFBSs to promoters may result in a misleading picture of 'promoter' content. We illustrate these issues using the example of histones H2A and H2B and show how such analysis could be misleading if care is not exercised to eliminate coding regions from the presumed promoter sequences.