CpG island prediction - (Jul/01/2008 )

Hi all,

I'm new to methylation studies and have been trying to understand the different CpG island prediction tools. I have been using http://cpgislands.usc.edu/ and http://www.ebi.ac.uk/Tools/emboss/cpgplot/index.html and the results overlap which is great. However, these don't appear to match up with the predictions provided in ucsc genome browser or map viewer (and these two don't match either). So where do i start?

Thanks in advance!

You know CpG island is an artificial concept and the term was invented to tell some regions in the genome are CpG rich. That is it. As far as biological significance is concerned, I think any size ranging from 100-500 bp is meaningful.

Here are some topics on the size of CpG island:
http://www.protocol-online.org/forums/inde...?showtopic=3203
http://www.protocol-online.org/forums/inde...?showtopic=3160

^^ lol at the above comment. CpG islands are "artificial" and "invented".....we have some good reason to believe that these regions are indeed discrete areas of chromatin, with histone modifications and proteins specifically enriched at these sites. will just agree to disagree methinks

As for the not lining up with the genome browser, it could be down to the tissues studied as metylation patterns are known to differ between cell types.

hi blue dingo,

the discrepancies you are seeing between the UCSC browser and other programs to determine CpG islands are most probably down to the way each program works in determining the observed to expected ratio and GC contents before they are flagged as a CpG island.

you really need to get into the scripting of the programs to see how each parameter is calculated.

I won't wager into the CpG island debate except to say they are guidelines and have been good guidelines for functional DNA sequences that can regulate gene expression, and there are a increasing number of exceptions to the rule!

Nick