Where to look for normal data when using sesame for CNV analysis on EPIC v2?
SixPlusSeven opened this issue · comments
Hi,
In many tumors, CNV is a crucial prognostic indicator. I'm planning to conduct CNV analysis on EPIC v2 data using sesame.
Could you please guide me on where to find normal data for use as a control? If public normal data is not available, is it permissible to profile EPIC v2 data from healthy individuals? What would be the minimum number of samples needed, and are there any specific requirements for the number of males and females?
best，
Alex
Hi Alex,
In our paper below, we used these EPICv2 data from the
https://link.springer.com/article/10.1186/s43682-023-00021-5/figures/6
This was done using the following two public datasets on the GM12878 lymphoblastoid cell line as control
GSM7139626 GM12878_rep1 [206909630042_R08C01]
GSM7139627 GM12878_rep2 [206909630040_R03C01]
Based on my experience, as long as the controls do not have strong copy number variation it will work fine. But it remains to check whether methylation level may alter results. I guess is it will be minor. There are more samples you may use in that GSE228820 for controls
Sex will affect sex chromosome copy number inference (obviously). If you use a male as control, and your female samples will show loss of chrY and gain of chrX. I wouldn't mix male and females in your control at least since that will blur the picture. Just two cents.
Thank you for your reply! I am doing the CNV test now.
In addition, I would like to ask you a question, does the sdf file need to be quality corrected, or just use readIDATpair or openSesame(,func = NULL,)?
Btw, these datasets will be made to sesameData in the next release (currently on devel).
Re normalization: I would say yes, particularly the dyebias correction might help. So openSesame(, func=NULL) will do. Based on my experience, it doesn't matter much in practice even if you don't and just do openSesame(prep="", func=NULL).