Methylation Study

[EpiBrass]

Hi,

I'm a PhD student and currently working on building up a Methylome of a plant species. I have a plant line which has been inbred for 9 generations to ensure homozygosity. I would like to know whether it is more robust to perform the sequencing from a) tissue of several individuals from the same mother pooled together and sequenced several times, b) tissue of several individuals from the same mother indexed then pooled, c) tissue of several individuals sequenced once each separately or d) tissue from a single individual sequenced several times?

* Using Illumina MiSeq *

Cheers,

Justin

[dpryan]

Is the goal simply methylation profiling or do you also want to look at methylation differences (say between two related inbred strains with different phenotypes)?

[EpiBrass]

Is the goal simply methylation profiling or do you also want to look at methylation differences (say between two related inbred strains with different phenotypes)?

The aim of this is to develop a standard methylation profile with which I can later use as a reference to compare back to for methylation studies using different tissues at different time points under different growing conditions using the same inbred line. Thus deducing which genes/genome regions are at least in part regulated by methylation modulation in response to developmental or environmental stimuli.

[SNPsaurus]

d) will tell you about technical variability, but I suspect that biological variation will dominate, so c) is the most robust (b is similar). If you know the variation, then you can assess if new samples fall within that variation or not. If you just do a) you will have the average, but you won't know if new samples differing from the average are examples of new methylation patterns or at the tail end of the distribution already seen.

[EpiBrass]

d) will tell you about technical variability, but I suspect that biological variation will dominate, so c) is the most robust (b is similar). If you know the variation, then you can assess if new samples fall within that variation or not. If you just do a) you will have the average, but you won't know if new samples differing from the average are examples of new methylation patterns or at the tail end of the distribution already seen.

Thanks for that. Definitely a helpful clarification! I think what I'll do is a couple of technical reps (a) as well as assess biological variation (c) from a couple of individuals.

[dpryan]

I agree with SNPsaurus, either C or B (if the genome is quite small) will be your best bet. If the genome isn't smallish, you'll need technical replicates as well to get enough depth (if you're planning bisulfite-sequencing, keep in mind that the alignment rate is a bit lower than what you would otherwise get). Having said that, you'll be better served sequencing more samples at (relatively) low depth than just sequencing one pool or one sample at high depth (estimating variability and all that).