1
$\begingroup$

For Bisulfite Sequencing data, we can calculate methylation level for a specific locus, $$ \mbox{methylation level} = \frac{\mbox{# methelyated reads}}{\mbox{# total reads}} $$ Here I think, for this locus, it can only have one status, methelyated or not. Or to say, methylation level here can only be 0 or 1. Why does it have other values in (0,1)?

$\endgroup$
6
$\begingroup$

You are right that a single molecule in a single position is either methylated or not methylated. However:

First, assuming your organism of interest is diploid (or of higher ploidy) one of the chromosomes could be methylated, the other not. That would give you a level of 0.5 and can be found in imprinted regions (where the paternally inherited chromosome is methylated and maternal not, or vice versa).

Second, except if you use a pure cell population of a single cell type it is likely that there will be cellular heterogeneity in your sample, e.g. from different types of blood cells or neurons vs glial cells in the brain. Some of these may have methylated CpGs in this locus (depending on cell-type-specific gene activity) while others can be unmethylated. Depending on the relative proportions of these cell types the methylation level can be anything between 0 and 1 for a mixture of cell types.

Thirdly, sometimes CpGs get grouped together. For example, 5 nearby CpGs are combined to give a single methylation level for this locus, and then the level can be 0; 0.2; 0.4; 0.6; 0.8 and 1.

Lastly, there could be technical artifacts or technical reasons why a position was not called as methylated (insufficient quality for a single read?), leading to deviation from 0 or 1.

$\endgroup$
2
  • $\begingroup$ Thank you very much! I noticed some databases supply a single methylation level for each promoter region, which I could not completely understand before. Now I think that's the third case in your answer. Your answer is really helpful. $\endgroup$
    – Jie
    Aug 5 '20 at 13:21
  • $\begingroup$ Thanks for the explaination. One further question though, why would the methylation distribution are biomodal when 1) methylation could be heterozygous and 2) we have mixture of cell type where each type represent different methylation pattern? $\endgroup$
    – unicorn
    Sep 16 '20 at 2:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.