I am using CHAMP tool to identify differentially methylated probes (DMP).
initially, after normalization I got methylation values between 0 to 1. What cut off value I should set to classify probes into
1) hypo methylated
2) hyper methylated
3) Un methylated.
Secondly, after getting DMPs between two groups (Normal vs Tumor), there is a delta beta value ranging between -1 to 1. How can I interpret this delta beta value?
Regards,
You mention values between 0 and 1, which probably is your beta values. A methylated CpG has a value of 1 and unmethylated 0. However, the beta values originate from intensity measures and are not always 1 and 0, but maybe more like 1-0.8 and 0-0.2 if you have a very homogeneous group of cells/tissue. If you are looking at DNA that originates from blood then this is mixture of many different cell types than e.g. "dilute" 0.8 down to a lower number. Consequently, you will have to look at your data in order to define hypo and hyper-methylated sites. I think from above you can figure out what the delta beta value is in your test that identifies DMPs.
Kind regards, Kristian
Thanks alot for the reply. Actually my question is as follows:
Forexampe, I have the following DMPs
- ProbeID Primary_AVG Metastasis_AVG deltaBeta
cg25537217 0.565909837 0.10111869 -0.464791147
cg10445453 0.477970543 0.046390185 -0.431580359
cg27066284 0.499521113 0.076068824 -0.423452289
cg17276590 0.539110628 0.117784147 -0.421326481
cg03035213 0.520186911 0.108597259 -0.411589653
cg16295461 0.460798358 0.846421978 0.38562362
cg16539957 0.23412924 0.615751367 0.381622127
cg05529816 0.222409101 0.592327353 0.369918252
cg01678309 0.416992146 0.784389964 0.367397818
what does "delta beta value -0.464791147" mean? As per my understanding, Primary_AVG group and Metastasis_AVG group both showed methylation .So, what I can understand from the deltabeta value?
Kindly suggest,
Regards,
The terms hyper and hypo methylation are typically used and meaningful as considered in relation to a control. Viewed in isolation however, in my opinion >0.8 would be considered hyper and
I convert the beta values from our Methyl450k arrays to a relative percent methylation, comparing to that of a "normal" control sample set (consistent with Dr. Karpf's "relation to a control"). There are a variety of normal thresholds, but I have seen thresholds as low as 5% when accompanied with FDR-adjustment, though 10% seems to be a more common threshold.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS