Quantitative ChIP-Seq: Normalization of Library?

[jazz710]

Hello all,

I'm posting today about some basic ChIP-seq experimental design thoughts. The goal of the experiment is to quantitatively compare the genomic occupancy of a histone variant in developing vertebrate embryos across development stages and across different rearing conditions.

To briefly explain my experiment: I have 24 IP samples (+24 input samples) which represent sampling of vertebrate embryos from two species, two treatments, and each comparing three developmental embryonic stages each with two biological replicates. (2 species*2 treatments *3 stages*2 replicates=24)

I want to make the following quantitative measures of histone occupancy:

1) Within species and treatment, across developmental stages
2) Within species and developmental stage, across treatments
3) Within developmental stage and treatment, across species

I would like for these analysis to be quantitative which is why I have included an input sample for each IP sample.

Currently, I am doing a 10% input method (ie. I do my IP on 1000ul of chromatin, and my input on 100ul). The goal (as I'm using the Illumina TruSeq ChIP-seq kit) is to get between 5-10ng of ChIP DNA per IP and to proceed onto analysis (currently planning to use Bowtie/MACS). As I actually do the IP reactions, I'm finding (not surprisingly) that some stages produce more IP product than others as the samples are physically different sizes. The input samples always produce plenty of DNA (>70ng per sample)

Finally, my question:

For truly quantitative ChIP-seq which is more important?

1) Always prepping your IP and input in the same ratio (ie. 10% input)
2) Adding the same amount of IP DNA and Input DNA to the library prep?

My gut tells me that Option 2 is more important. If I do my library prep with 10ng of Input (from 1 single Input DNA extraction) and 10ng of IP (pooled from multiple IPs) that is still appropriate for quantitative analysis, right?

Thanks for your help with this! I'm also going to chat to some folks at my University that do ChIP-seq, but I figure I'd ask here too since there's such a great knowledgebase here.

Best,
Bob

[jazz710]

After talking with folks (including Illumina), Option 2 is appropriate. Matching your input and IP samples loaded into library prep gives you quantifiability.

I had the 10% input idea in my head because my protocol is derived from a qPCR ChIP assay where that was integral in comparisons between samples.