I'm struggling to make sense of my ChIP-Chip visual data. Let me precede all of this by saying I've done numerous controls and tests at the sonication, IP, and array steps, all of which appear to give me accurate data. However, when I visualize the final ChIP-Chip data, it appears that there are too many jagged peaks throughout the genome, including some large, wide positive and negative peaks. Still, there are also some textbook ChIP-Chip peaks there as well.
But here's the kicker - I've done three completely separate biological replicates from culture to array and get the same results, i.e., similar peaks between each array (even those positive and negative peaks that don't look quite right). So if the data is repeatable, is it really noise? Or is it indicative of actual binding? I have read that noise can contaminate an array data set and give false peaks, but also that there is the 'appearance' of noise when evaluating global repressors, and that it's actually correct data.
Specs: I'm evaluating a repressor protein in E coli using Nimblegen K-12 WG microarrays.
I don't have the experience with ChIP-Chip data to make final conclusions about this data, so any help from any of you who may have seen this sort of thing before is greatly appreciated. I've attached a small (~90kbp segment of the genome) pic of the three replicates so you can see the repeatability. Note that this is raw data, before the peak-finding algorithm is applied.
Edited by chrismbyrd, 19 October 2009 - 10:58 AM.