Analysis of Quantititive Sequencing - (Jul/18/2007 )
Hi Guys,
I'm trying to do the data analysis of my pyrosequencing results and am having some difficulty determining the best way to represent my data with relation to age or parity. I have roughly 50 samples which have each been analyzed for 6 genes with each gene having between 8-12 CpGs. After doing some reading in the literature, I see that a lot of people tend to express the % methylation of the promotor as a single value known as the methylation index. What I'm having trouble figuring out is how this value is calculated.
If for example I had 1 sample which had methylation values of :
4%, 6%, 8%, 5%, 7%, 8%, 3%, and 6%
CpGs analyzed = 8
mean methylation = 5.88%
How would I go about calculating the MTI?
Also, as I'm looking at multiple genes, what would you recommend as the best way to represent all of the data on a single graph? I had thought to use the average Z-score, but I'm not sure if there would perhaps be a better way to do this.
Any help would really be appreciated!! I'm terrible with stats!
4%, 6%, 8%, 5%, 7%, 8%, 3%, and 6%
CpGs analyzed = 8
mean methylation = 5.88%
How would I go about calculating the MTI?
To my knowledge there is no common rule for this. The MTI has been introduced to account for imprecise measurements by transforming the results into a 4 or 5 step scale, e.g. 0-20% = 0 21-40% = 1 etc.
Given the high sensibility of pyrosequencing, I would use the raw values. An average value for the whole promoter can be misleading, as you might have changes in only some of the CpGs.
If you use the average amount of promoter methylation, I would plot the %value and it's SD per gene and group. This will be easier to interpret than trasnformed Z-scores.
Hope that helps,
Krümel
"An average value for the whole promoter can be misleading, as you might have changes in only some of the CpGs."
Hi Krümel,
Its interesting to me that you mention this. I've noticed in some samples that some CpGs in the genes I'm examining seem to exhibit slightly higher levels of methylation in relation to age. I had thought this might be some sort of an artifact, but these trends seem to be reproducible as I have resquenced a number of these samples and gotten the same results. In talking with Issa at the Annual american cancer research meeting in LA. He seemed to suggest that when methylation occurs it occurs in a very uniform fashion across the CpG island which I believe is why he plots averages of the CpGs across the island and z-score averages in a number of his papers. He works with prostate tissue and I work with breast tissue, so I dont know if this may be a tissue specific phenomenon, but I found it interesting nonetheless.
I think that it might be a gene-specific phenomenon. There are examples for non-uniform methylation of CpG sites. Your observation about age and methylation is very interesting, if I were you I would follow that path a little... you never know 
Are these examples in human tissue? And would you happen to know any references off hand krümel? Thanks
Just one example at hand: PMID: 17207969
Krümel