Hi,

I have no experience with the MeDIP protocol but there are some publications comparing the different methods:

Nat Biotechnol. 2010 Oct;28(10):1106-14. doi: 10.1038/nbt.1681.

Nat Biotechnol. 2010 Oct;28(10):1097-105. doi: 10.1038/nbt.1682.

Genome Res. 2010 Dec;20(12):1719-29. doi: 10.1101/gr.110601.110.

Epigenomics. 2010 August 1; 2(4): 587–598. doi:10.2217/epi.10.36.

Methods. 2010 Nov;52(3):203-12. doi: 10.1016/j.ymeth.2010.04.009.

It really depends on the fundamentals of your research hypothesis, ab for MeDIP works good nowadays. MBD2/MIRA could be much easier for down-stream analysis, as the data are cleaner.

What about RRBS instead of enrichment? (Reduced representation bisulfite sequencing for comparative high-resolution DNA methylation analysis, Alexander Meissner et al. 2005, Nuc Acid Res) It also reduces sequencing costs and may be an alternative?

First, think about what will You like to know from DNA methylation data. The link to gene expression depends strongly from the exact position of methylation in relation to the genes. So I would also prefer a bisulfite sequencing technique because You will get exact positions and absolute amounts.

Hi Phyllis,

This is a good question and there are a myriad of things to consider. Both MeDIP and

MIRA have inherent sequence biases. We have had difficulties with commercial MIRA kits and making the MBD proteins requires expertise in biochemistry. Therefore I would favour the MeDIP approach. There are many ways to render the ssDNA immunoprecipitated by meDIP to dsDNA [whole genome amplification included, although be cautious with the number of cycles used; less is more]. Also, we have used RRBS but the basic RRBS protocol is heavily biased toward CGI's and you will lose coverage at interesting enhancer elements and perhaps orphan CGI's which may be relevant in the developing brain.

The way I might approach your quandary, is to consider higher coverage RRBS which uses multiple restriction enzymes to fractionate the genome therefore covering more than just the MspI smear - and thus covers non-CGI and non-genic regions also. Best of luck with the project, Donncha.

Hi Phyllis

I am commercial and operational director at NXT-Dx. We are a company specialized in epigenetics and DNA methylation in particular. We offer services to our global customers in this field and both MBD-seq and MeDIP-seq are techniques which we offer.

Based on this I feel we are in a good position to answer your question. A lot of the things that have been said by other people in response to your question are true.

MBD-seq does focus more on CpG islands than MeDIP does. We know from experience that fragments containing around 4-5 CpGs are more likely to be picked up by MBD-seq. This is not the case for MeDIP.

However in terms of downstream analysis of the resulting data MeDIP-seq will be more difficult since it has much more background signal than MBD does. The data from MBD-seq is cleaner and easier to interpret.

Besides this it is important to realize there are several commercial kits available for doing MBD-seq, which do not react in the same way. We have tested all the kits available on the market and have seen important differences between the kits and the resulting data. I can refer you to a paper written by our principal scientist on this topic, if this would be of interest to you.

Also I would be happy to chat things through in more detail if you would like. If interested in a chat, please let me know.

My email address is [email protected]

Good luck.

Maarten

Hi,

I want just to emphasize that in case of MeDIP-seq the easiest way to prepare a library is to ligate the sequencing adaptors before doing the IP. The work flow would be the following: DNA isolation, DNA shearing, ligation of adaptors, DNA denaturation, IP, PCR amplification.

For details see

PMID: 22402632

PMID: 22569366

Best

Filippo

We do MeDIP. there are several important factors to bear in mind.

The anti 5-mC antibody works best on ssDNA. so either you put adapters on LARGE amounts of DNA pre-MeDIP, or you work on the ssDNA afterwards.

Initially, putting adapters on before MeDIP looks like a good idea, however, when you start working out the costs of adapters, ligation and nick repair on up-to 4 micg of input DNA, then you start worrying. We also found that home-brew was nowhere as reliable as the manufacturers ligation kits, which makes things more expensive, as most are optimised for 100ng Input DNA.

We have a protocol that takes the IP'd ssDNA, makes the 2nd strand, ligates the adapters, and gives you sequencing ready samples. There are now commercial ssDNA kits available for library preparation, so this is not a worry.

We looked at setting up MBD-seq, and couldn't finance the time necessary to set it up although there are now MBD-seq kits available commercially, however, I don't know what they are worth. Similarly RRBS was excluded, partly because of the sequencing requirements, and the shear volume of repetitive elements that are sequenced, that can not be uniquely aligned back to the genome (as well as the inherent difficulties in aligning bisulphite modified sequences back to the genome.)

Jon