SEQanswers

Go Back   SEQanswers > Applications Forums > Epigenetics



Similar Threads
Thread Thread Starter Forum Replies Last Post
Newbie question: low cluster density on MiSeq (Chip-Seq pool) HelenaSC Sample Prep / Library Generation 12 07-18-2013 07:28 AM
bwa question on file outputs/inputs ikrier Bioinformatics 12 10-12-2012 08:22 AM
question about random select certain number of reads from ChIP-seq bed file [email protected] Bioinformatics 1 09-15-2012 02:13 PM
Gene prediction with both RNA-seq mapping and genomic sequence based inputs? jstjohn Bioinformatics 3 04-08-2011 10:59 AM
Yeast ChIP-seq; Sonication/Bioanalyzer question AaronS Sample Prep / Library Generation 1 01-14-2011 06:55 AM

Reply
 
Thread Tools
Old 11-13-2013, 11:17 AM   #1
omy567
Junior Member
 
Location: CA

Join Date: Aug 2013
Posts: 3
Default Question about inputs for ChIP-seq

Hey guys. I just got ChIP-seq working, and I am gearing up to sequence some more libraries. I had a question for the forum. I am sequencing ChIP DNA from the same cell type that I have already sequenced input from over 2 different cross-links. Is it common practice to sequence an input from every crosslink you ChIP from, or can I just compare the sequence data from my new ChIP libraries to the input DNA that I have already sequenced from the same cell type? Is this accepted in the field? It has been suggested to me that I do this, but it seems strange to me not to have an input with every crosslink.

Thanks.
omy567 is offline   Reply With Quote
Old 11-13-2013, 02:30 PM   #2
Wallysb01
Senior Member
 
Location: San Francisco, CA

Join Date: Feb 2011
Posts: 286
Default

For each crosslinking you should sequence the input along with the ChIP. Experimentally, its best to keep everything in parallel as much as possible. The crosslinking and sheering are going to be different with each new input. For me, I like processing the inputs and ChIP libraries as much in parallel as possible beyond that. So, DNA isolation all the way through library prep and sequencing are done together. Though sometimes you end up with your input and some of your ChIP libraries (from the same crosslinking) being sequenced at different times due to machine problems, but you do the best you can.
Wallysb01 is offline   Reply With Quote
Old 11-13-2013, 02:49 PM   #3
jwfoley
Senior Member
 
Location: Stanford

Join Date: Jun 2009
Posts: 179
Default

Input isn't really a negative control (it measures chromatin accessibility, which is of course enriched where you find transcription factors) and arguably you shouldn't be doing any input samples. I certainly wouldn't recommend spending money to get more of it.

Is there a more biologically meaningful control you could do? Can you knock out/down your ChIP target? Do you have a different experimental treatment? A different antibody to the same target?
jwfoley is offline   Reply With Quote
Old 11-13-2013, 05:27 PM   #4
Wallysb01
Senior Member
 
Location: San Francisco, CA

Join Date: Feb 2011
Posts: 286
Default

I think what jwfoley is arguing for is some sort of biological control, while you might call inputs an experimental control. Both have validity, it just depends on what you're doing. You may want to do one or the other, or both. It sounds like you're already doing both, and I suggest you keep sequencing your inputs just incase some crosslinking/sheering differences creep in.

Here's what the encode best practices has to say, which can be found here and has tons more useful information for those relatively new to ChIP-seq https://genome.ucsc.edu/encode/proto...e_v2_2011.pdf:

Quote:
Controls
For both ChIP-seq and ChIP-chip, control experiments must be performed. Breakage during sonication can occur preferentially in regions of open chromatin resulting in non-uniformbackground signal (Auerbach et al., 2009). In addition, many cell lines are aneuploid and have many large regions of genomic duplications, which can heavily influence peak sizes and rankings. Control DNAs include “Input” DNA, in which DNA is isolated from cells that have been crosslinked and sonicated under conditions similar to the experimental sample or “IgG” in which control immunoprecipitations are performed using an antibody fraction (Immunoglobulin G from an unimmunized animal) that does not recognize DNA or chromatin associated proteins. In cases where factor binding to DNA is environmentally induced, (e.g. after hormone induction of the glucocorticoid receptor or when analyzing a protein expressed from an inducible promoter), a potential appropriate control is a parallel ChIP experiment done on cells in the uninduced conditions. Similarly, for epitope tagged constructs suitable controls are performed with cells lacking the epitope tag. If amplification is used to prepare the experimental sample, then the control DNA must be prepared using the same amplification procedure; note that biases in amplification can increase the chances of overrepresentation and underrepresentation of sequences. Control experiments need to be performed for each cell line, developmental stage and different condition/treatment since the open chromatin regions are likely to change under the different cell types/stages/conditions.
Wallysb01 is offline   Reply With Quote
Old 11-13-2013, 05:39 PM   #5
jwfoley
Senior Member
 
Location: Stanford

Join Date: Jun 2009
Posts: 179
Default

Quote:
I think what jwfoley is arguing for is some sort of biological control, while you might call inputs an experimental control.
Better to call input a technical control. But yes, if you do a biological control, a technical control may be redundant. And if your only control is technical then your biological findings might not be very solid.
jwfoley is offline   Reply With Quote
Old 11-13-2013, 06:32 PM   #6
Wallysb01
Senior Member
 
Location: San Francisco, CA

Join Date: Feb 2011
Posts: 286
Default

A single Chip is a single experiment. You take "A" DNA and pull down a subset "B". "A" is then the control of that singular experiment. The wider experiment may contain different stages, treatments, tissues, antibodies, etc., but they all hinge on the simple put A in get B out. Then you can start comparing B1, B2 ... Bn when correcting each for A1, A2, ... An. Arguig over technical control or experimental control is largely an argument over semantics.

My assumption is that omy567 has A1, A2 and B1, B2, but is now wondering if he can just continue with only B3 ... Bn. I would argue that's not very wise as it relaying on the assumption that cells grown separately, cross linked separately, sheered separately, run through library prep and sequencing separately will yield similar data.

You only need maybe 1/4-1/6 of a 1x50bp HiSeq lane, which could cost as little as $100-200 in library prep and $200 sequencing. That's not a lot of money to protect yourself from batch biases.
Wallysb01 is offline   Reply With Quote
Old 02-12-2014, 08:59 AM   #7
zwang0614
Member
 
Location: uk

Join Date: Feb 2014
Posts: 10
Default

The Input is better to be included for each chromatin extraction. There are quite a lot of reasons to do so.

http://nar.oxfordjournals.org/content/39/15/e103.long
zwang0614 is offline   Reply With Quote
Old 02-12-2014, 09:27 AM   #8
jwfoley
Senior Member
 
Location: Stanford

Join Date: Jun 2009
Posts: 179
Default

That paper says exactly the opposite of what you're saying.

Quote:
We show that relying on input control as a normalizer is not generally appropriate
In particular:

Quote:
However, we observed that C. elegans ChIP input control DNA sequence data (hereafter referred to as input sequence) display strong reproducible patterns (Figure 1a). These patterns do not appear to be due to cross-linking and incomplete solubilization of DNA fragments in the input extract because similar patterns are evident in published genomic DNA sequence data (7), where the DNA was deproteinated prior to fragmentation and not cross-linked (Figure 1a). The reproducibility of the patterns indicates that they are not due to random sampling variations.
In other words, not only do you not need to do an input control for each ChIP, but you probably don't need to do an input control at all, because the biases it captures are constant across all experiments on the same genome. The biases are inherent and biological, not technical.
jwfoley is offline   Reply With Quote
Old 03-20-2014, 07:58 AM   #9
dbrg77
Junior Member
 
Location: Manchester

Join Date: Sep 2010
Posts: 6
Default

It is better to include input for every chip you do so that you keep everything in parallel.

Having said that, we have found that peak calling results have negligible differences using different input as long as it is from the same cell type. Now as long as we have one input for a particular cell line, we do not do input at all.
dbrg77 is offline   Reply With Quote
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off




All times are GMT -8. The time now is 08:24 AM.


Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2018, vBulletin Solutions, Inc.
Single Sign On provided by vBSSO