Sorry, but I'm not using IPA.

lol...thanks Daniel - they have a good customer support but I've just been reviewing literature to see innovative ways other researchers have used the software to publish data - thought i'd ask around here as well :)

Sorry, but I'm not using IPA.

That's alright i guess very few people are. Thank you

You may stratify those CpG by gene. Choose the genes containing DMR in promoter region into IPA analysis.

Thanks Edward! That's a useful tip. I've come across a lot of papers that do that and then make venn diagrams to show overlapping networks/functions common to the hypo- and hyper-methylated genes to re-enforce the conclusion.

Dear Yudhishtar (I like your epic name very much),

I also used IPA in similar way as it was mentioned by you and Edward. However, I also add the gene expression based on publically available datasets. That means you have genes not only significantly methylated DMRs between your groups, but also significantly expressed as well. But then you need expression data as well and I used the direction of expression data in the end which is more reliable. If you need further details or more info don't hesitate to contact me further.

Goodluck,

Tushar

Haha - thanks Tushar, my mom was fascinated by the character in Mahabharat :P

Could you give me names or examples of a few of these publicly available datasets? If i upload my DMR information would it return accurate gene expression data? I'm highly intrigued by this, as of now I'm simply assuming the genes whose promoters are differentially methylated are either expressed or repressed depending on the met state but I'm unable to put an exact value and I can't be sure that the hypomet gene promoters would be over expressed simply because they are more accessible for transcription.

Thanks for your answer! :D

Hi Smeeta,

I think Tushar might have a bit more expertise in this than I do. His inputs were very valuable. I hope I'm not getting it wrong but I believe he was suggesting that we look at publicly available datasets where cpg methylation is correlated with gene expression or not as sometimes hyper or hypomethylation may not definitively imply transcriptional repression or expression. In my case I have data from a study done some time ago and the person who did the analyses for CpG methylation used some softwares and possibly databases like Tushar was talking about to find out which gene promoters were involved in this and sent it to me. Using that data, I'm adding those genes to IPA (which I've gotten pretty good at using) and checking for disease, function and canonical pathway networks using the core analysis. Also microRNA target filters are very useful if you have some miR data (in our case we do). But mostly my query was how to use the hypo vs hyper methylation information. Thats a bit tricky in IPA since they have not integrated with the Illumina450K array yet or any other methylation based arrays as far as I know. The hyper or hypo may be interpreted as silent or active (more confidence in hyper being silent) marks but you can only use that in IPA by either using Molecule Activity Predictor from "overlay" in your pathway or you ascertain up/downregulation in your entry dataset of genes.

Hope that helps you a bit using IPA. As for interpreting 450K data I'm still a bit wet behind the ears so maybe dig a bit deeper in that area :)