Skip to content

MeDIP-seq protocol results

March 2, 2012

I finally got the data back from my MeDIP-seq experiment.  After months of waiting for everything and anything (Special thanks to all our ‘wonderful’ Greek distributors of scientific products).  Enough to make me pull my hair out and make my collaborator Andrija wonder if he’ll ever get his Ph.D.  Anyway, the protocol itself seems to work really well.  Below are a couple USCS genome browser screen shots. The first is of a methylated locus and you see a nice peak in the middle of all the tracks.  The second is the GAPDH promoter and there is no signal despite the high GC content.  So if you are looking for a good protocol for making TruSeq compatible/barcoded MeDIP-seq libaries it is here:  Barcoded MeDIP-seq protocol with TruSeq adapters.

Now on to the data analysis, where the tools appear to still be a little rough around the edges. And most importantly of all hoping that there is some biologically interesting data to get out of this experiment.


From → Posts

  1. Hah I hope you ran a fully methylated control as one of your samples! Makes he data a lot easier to interpret.

  2. ethanomics permalink

    Hi Aaron,
    No, I sure didn’t but I wish I had!!! No experiment goes perfectly the first time. Do you use SssI DNA methyltransferase to methylate your genomic DNA?

    This was just a pilot experiment anyway, to see if the method was working and I am really only interested in differentially methylated regions so each condition serves as the base-line for the other condition. So I am not a total idiot. I was going to run input DNA, but it didn’t really make any sense and it had the same bar-codes as my immunoprecipitated DNA, so I left it out. Fully methylated immunoprecipitated DNA, on the other hand, makes a lot of sense. Thanks for the tip!!!

    Which leads me to wonder about what to use for ES cells, which have significant levels of non-CpG methylation. Do some tumor cells have higher levels of non-CpG methylation?


  3. The screenshot from UCSC looks very nice and it seems you do some kinds of smoothing. How dou you smooth your data? MACS?

  4. ethanomics permalink

    I agree they do looked smoothed but if they were I didn’t do it. What I did was map the fastq files with bowtie, convert the output sam files to bam files and sort them with samtools, convert the bam files to bed files and then bedGraph files with bedtools. There is no option for smoothing as far as I can see in that pipeline. Maybe UCSC automatically smoothes bedGraph data?

    • Thank you for your replying. I am fellowing your pipeline to check it. I usually visualize my data by wiggle file

  5. ethanomics permalink

    Actually, I wasn’t completely accurate with my response. If you download the script ezDESeq which is linked in the scripts tab above it will also download a script called ezMACS. There is one extra step of processing. I extend the read length to the average size of the inserts before I convert the bed files to bedGraph files.

    One thing I did not do which just added to my data analysis pipeline is to normalize by total read count. In this case the read counts were pretty much equal so it didn’t matter, but it’s a good idea to do.

    • Do you mean that you use MACS to call peaks? I think that MACS can smooth data for wiggle generated by MACS looks smoothed

      • ethanomics permalink

        I’m not using MACS to call the peaks, just that script to make the bedGraph files. MACS did not find any differences between my two conditions. MEDIPS seems to be identifying differentially methylated regions that are real, but it also outputs a lot of false positives.

  6. ethanomics permalink

    One other little bit of information that may be of help. Some scripts that convert files to bedGraph files reduce the resolution to decrease the file size. This makes them look less smooth and a little uglier.

    • I generated bedGraphy using my own data and fellowing your pipeline, It looks like what you show in UCSC Genome browser. But I didn`t see smoothing by wiggle file even 10bp bin which I think is very similar with bedGraphy (bin in bedGraphy is not constant but mean of bin size is about 10bp)

Leave a Reply

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

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

Google+ photo

You are commenting using your Google+ 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 )


Connecting to %s

%d bloggers like this: