This is a general question about results that I have seen several times but only recently considered. If you do something like ChIP-Seq or HITS-CLIP or other techniques that immunoprecipitate DNA or RNA bound by a protein, then run motif analysis on the areas of the genome mapped to the reads that are pulled down, but do not find the binding motif for the protein that you used for the IP, what does it mean? Is this expected? Did something go wrong with the wet lab steps, or analysis?
In my opinion it suggests one of two things: 1. Either your antibody has specificity issues. It is not uncommon that antibodies are pulling down other areas than the expected. TSS in particular. Best control for that is to look for signal in a knockout cell line. I know some people who raised five (as I remember it) antibodies against a target and got quite consistent peaksets in chipseq. They later discovered that all antibodies resulted in plenty of peaks in a cell line that did not have the gene encoding the target, suggesting that the peaks were due to some crossreactivity or 'sticky' regions. Ouch... Not a good day in lab. 2. Alternatively the antibody recognizes the target properly, but the assumed motif is not right or the majority of binding is due to indirect recruitment. Again, data from a knock out cell line would be handy to convince your audience of that.