The identification of cis-regulatory elements responsible for transcriptional and translational control is a major challenge in the understanding of gene regulation. The occurrence of SNP in cis- elements might profoundly modulate gene regulation. Classical methods involve matrix techniques for the prediction of known regulatory elements. However, these methods can only applied if binding matrices are known. Experimental approaches use deletions variants which may insufficient e.g. if shortening leads to changes in the secondary structure, an important feature for cis- element/ trans-factor interaction, especially in RNAs.

Here, we propose a combined experimental and theoretical approach to predict cis-sites for RNA/ protein interaction in their native secondary structure. This approach is based on UV crosslinking data of RNA/ protein interaction with different radioactive labeled nucleotides. The label-transfer technique allows estimating the relative amount of each possible nucleotide, involved in the direct RNA/ protein interaction site. The relative amounts r[A,C,G,U] of UV cross-linking signals are then analysed theoretically. An error function, comparing predicted signals and measured signals is used to calculate a relative binding probability along the given RNA sequence. We show experimentally that a site, predicted by our method, is of functional importance for gene regulation.