Skip to content

ChIP-seq with TruSeq adapters results

August 19, 2011

I got the data back from the ChIP-seq samples I sent out using my TruSeq library preparation protocol. I can definitely say the library preparation protocol works great, clearly better than what Illumina offers (less PCR cycles). 93% of the reads were unique.  There doesn’t seem to be any obvious GC-bias visible by eyeballing it. 73% of the reads mapped which I think is pretty normal for 50 bp reads on the human genome. There may have been some adapter dimer in the sample but I still have to look into it.  The only thing that was not good was that I only got 50 million reads total which was enough because I only pooled two samples, but still annoying in that the HiSeq 2000 is suppose to produce upwards to 150 million reads/lane. Since I do not have anything to do with the sample after the library preparation, I don’t know where the variability is coming in here. Also of note, the samples were sheared with the Covaris instrument and produced a nice even distribution of input tags with no major peaks at transcriptional start sites or other areas of open chromatin. All in all the methodology seems to be working really well. I think if you want to barcode ChIP-seq samples for the Illumina HiSeq 2000 this is the way to go. Again the protocol is here: ChIP-Seq TruSeq Library Preparation Protocol

Below are a couple of UCSC genome browser screen shots showing the input sample at a couple random spots in the genome, one of higher GC content and the other of lower GC content.

Advertisements

From → Posts

10 Comments
  1. Tony permalink

    We’ve seen the reduced numbers of reads before. It seems to be more common on IP samples than RNA and/or genomic DNA. I can only assume that the problem arises from having a reduced PCR protocol. More PCR cycles means the proportion of successfully ligated fragments in the library increase and more accurate library concentration when using the Bioanalyser. Of course, increasing the number of cycles leads to more PCR duplicates. The best way around this is to switch to a qPCR method of library quantification. We use the Kapa kit, although you could probably fashion a homebrew qPCR using standard 2X SYBR mix and the same PCR primer cocktail. You’d need some already sequenced samples at known concentrations determined by sequencing to use as standards.

  2. ethanomics permalink

    I’m using the Kapa Library Quantification Kit to quantify my libraries (but thanks for the suggestion). While it is better then quantifying with the Bioanalyzer, I think there are still some issues. I have heard that it helps to have a standard that is the similar to the library you want to quantify, e.g. use a human RNA-seq library standard to quantify you human RNA-seq libraries.

    The reduced number of PCR cycles is because the Kapa HF Polymerase I’m using for the library amplification requires fewer cycles to reach the same level of amplification then is needed with Phusion. With the exponential nature of PCR it is a very fine line between under-amplification and over-amplification. I think next time around I’ll order the Kapa HF Library Amplification polymerase with SYBR Green so I can monitor the amplification in real time and take them out exactly when the amplification is sufficient but not over.

  3. ChIPSeqNewbie permalink

    Hi Ethan,

    we are about to start our own chip-seq experiment and will try your protocol. Have you had the chance to try clean-up with Ampure XP beads instead of columns yet? Are you still happy with the results of your chip-seq experiments or do you have any further recommendations on how to improve the outcome?
    Thanks a lot already for sharing your protocol and also for any further helpful comments from your side.

  4. ethanomics permalink

    The library preparation protocol is working great. I haven’t seen anything published or sold that I would consider using. My only complaint is that it is pretty labor intensive. You can do the whole thing in a day if you are efficient, but it is a long day. I haven’t tried the Ampure XP beads mostly because I don’t think it would improve the results. It would be fun to see if you could get rid of the gel extraction step, but that would take some optimization.

    Good Luck!!!

  5. Dave permalink

    Hey Ethan-omics,

    Very interesting protocol and I think I might give this a shot. I have done loads of ChIP-Seq before but now I want to do multiple TF ChIP-Seq and would like to reduce the costs substantially by running them in as few lanes as possible. My question is how many ChIP samples have you multiplexed in a single lane to date and how many reads/sample did you end up with? I recently pooled 6 RNA-Seq reads in a single lane and got a total of about 250 million (50 bp) reads, so I guess the more samples you throw in there the better I guess…..up to a point of course.

    Thanks!

  6. ethanomics permalink

    Hi Dave,
    We are doing mostly histone ChIPs, which requires more sequence then transcription factor ChIP do to the increased number of biding sites. It also depends a lot on the specificity of your antibody. But we have pooled up to ten samples and got back some good data. The limit is always the number of ChIPs. If the sequencing depth isn’t enough you can always sequence the same sample again. 250 million is a lot of reads for one lane on the HiSeq. I have never gotten back that many. I would think that there would be a big hit on quality up that high, but as I said I have never gotten that many reads.

    A bunch of people have used the protocol successfully so I would give it a go. But, try it once with some input DNA. It would be no big deal to lose 10 ng of input DNA but losing 12 ChIPs would be a whole different story!!

    Ethan

  7. Mike permalink

    Hi Ethan,

    Could you write with what kind of cells do you work (big/small genome)? We are planning to do multiplex sequencing of ChIP’d chromatin with antibodies against H3K4me3, H3K27me3 and H3K27ac. As far as I know it is not optimal for ChIP with histones and human genome but we are tight on cash at the moment. Do you think that pooling these 3 samples and input tagged with different barcodes could give reliable data?

    BW,

    Mike

  8. ethanomics permalink

    Hi Mike,
    I’ve done ChIP with human and mouse cells. A good HiSeq run will give you >160 million reads per lane. If you 4-plexed (mixed four samples in one lane) that would still give you 40 million reads per sample. That’s enough for histone modifications with a lot of binding sites. You could even get away with less reads for preliminary experiments and go back and sequence the same samples if you needed more reads. This is all opinion and you might want to get some others.

    Ethan

    • Mike permalink

      Hi Ethan,

      thanks a lot for a reply. The problem is that we are using “Genome Analyzer IIx” which theoretically can get 40 million reads per lane bat in practice for ChIPseq 12 million is a good result as far as I know. So what would be your opinion right now? What would be the minimal number of reads accaptable for ChIPseq with mentioned above antibodies?

      Have a nice saturday,

      Mike

      • Hi Mike,
        I think the best practical advice is to multiplex your samples. Sequence them and if you need more reads sequence them again. You can just resequence the same libraries as you’ll have enough library to sequence them many times. Even if you need more then one lane, multiplexing reduces bias so you should multiplex regardless. There are some publications on this.

        As far as the GAIIx, someone needs to tell the people making the decisions at your institution/department that they are wasting time and money hanging on to a GAIIx. Worse yet, they are making it impossible to do good science. From what I understand (this is just what I’ve heard – I do not have the pricing on this) the cost to sequence a flow cell on a HiSeq is about the same as a GAIIx but you get >4 times more data from HiSeq. Tell them to mothball the GAIIx, find a good core facility and get your samples sequenced on a HiSeq.

        I know what is going on, someone spent a lot of money on the GAIIx and they feel they need to use it so they don’t feel like an idiot who wasted a lot of money. This has happened so many times over around the world it is a shame. But someone needs to end your departments downward spiral of waste and the sooner someone faces the truth the better your science will be. The GAIIx belongs in a museum, the HiSeq will belong there before we can blink. Sorry for the rant but I understand your situation and it frustrates me.

        Good luck,
        Ethan

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: