Sibel- it sounds like there is more than one issue here to troubleshoot. In general, I think it is more meaningful to normalize your data to input rather than background in terms of enrichment. Interpreting whether the degree of enrichment relative to input is biological significant is another matter. For that purpose, the best control is to design qPCR primers to adjacent virus elements and/or host genomic sites where you don't expect strong signal for your histone mark of interest. This comparison sets some benchmarks as to whether the enrichments you observe really stand out or not (analogous to whether you'd really see a peak or not in a genome browser type view of a ChIP-Seq dataset).

But perhaps your most pressing issue is signal specificity of your qPCR oligos. I'd suggest several controls here. First, have you run an in silico PCR test with your oligos and host genome? If you are working with retroviruses, there can be numerous retroviral sequences resident in your host genome. Perhaps your qPCR oligos recognize these?

Second, you can do some simple analyses to figure out what is being amplified in your mock reaction. Run a denaturation curve to see if you are getting a unique DNA species amplified, and also run an agarose gel to assess if the amplified material is the expected size. You might be observing some non-specific priming, or if you have a unique species of the expected size, this result would indicae contamination of your virus in some stock reagent or consumable (tips/plates) in your assay. These tests will help point the way on the source of your signal in the mock samples.

Great! So, i am working with adenoviruses-dont integrate to host genome-. I was also thinking to run same assay on some cellular genes such as 18S. But my doubt mainly is about normalizing the data to input without taking background (beads) signals into consideration. I know that the viral copy number changes between the days of infection which is contributing to background signal. Because, viral DNA might get stuck on the beads and I personally think one should subtract the background signal. How about subtracting background signals first and then normalize the data to input? Thanks a lot!

Sibel- I still think you should invest some effort eliminating that spurious signal in your mock infected samples. Once that issue is resolved, you can tackle the normalization. It will be laborious, but you really want to run input for every sample/condition, and normalize internally to those. To be explicit, enrichment in time point 1 relative to input time point 1, enrichment in time point 2 relative to input time point 2, etc. This will correct for changes in copy number.

Beads alone or a control IgG are useful negative controls as well, but once you have conditions worked out for the ChIP where these levels are low, I don't think you need these negative control conditions every time you run the assay. And I wouldn't necessarily plan on using them for normalization purposes.

I agree with Daniel's point that if the mock sample is truly uninfected, there should be no amplification at all with your viral-specific promoter primer pairs--and this must be resolved first before attempting any analysis. A gel analysis of the products from the "mock" sample and the infected samples could help you determine whether you are dealing with contamination or "non-specific" amplification. Just reviewing the melt curve analysis--if you are doing Sybr qPCR--should also be helpful. One important tip. If you are using a Sybr mix that contains UNG--it is essential that you run the qPCR products on an gel immediately (IMMEDIATELY!--have the gel ready to go) after the PCR run is finished. Otherwise, the UNG can "renature" and start chewing up your newly formed products and you will only see a smear when you get around to running the gel.