Will over-amplification during library preparation "drown out" ChIPseq peaks?

[JIrish]

I'm in need of some help in order to move forward:

I'm a grad student in a lab brand new to next-gen sequencing, I'm trying to run ChIPseq experiments with ER alpha Ab (sc543) on an ER+ breast cancer cell line, and I've now tried to analyze two sequencing runs on two samples, with the analysis workflow summarized below.


My question: could over-amplification during library preparation cause the peaks in my ChIP sample to be attenuated, or is it more likely that my antibody just didn't enrich specific genomic regions? Both input and ChIP samples were amplified 18 cycles.

Illumina Hi-Scan > bowtie2 > SortSam > macs14 (ChIP and Input) > ~1100 peaks with FDR range of 75-100

.wig files for treat and control generated by MACS have almost identical profiles for any given chromosome when viewed with IGV.

[kopi-o]

Hard to say without seeing the data but as a rule, over-amplification artifacts can often be picked out by eye as "boxy" regions when looking at the coverage in a genome browser like IGV or UCSC. If the wig files have almost identical profiles for control and treatment then I would lean towards thinking that the antibody didn't enrich specific regions.

[Dario1984]

.wig files for treat and control generated by MACS have almost identical profiles for any given chromosome when viewed with IGV.

A better way to understand if you have amplified too much is to look at the peaks. Do the reads start from a variety of positions ? Or, do they mostly start at very few positions and are there sizeable gaps between them ? If the reads are spread out well, but low, then your antibody didn't work well. If you have peaks with gaps of no reads between positions of high reads, then you have amplified too much.

Some experienced researchers have made a protocol to determine what is the best number of PCR cycles for a particular sample.

[JIrish]

Thank you both, I'll get an IGV screenshot up here soon if someone wouldn't mind taking a look; I suspect the antibody just isn't working, but it would be nice to know if over-amplification is a contributing factor.