Methylation - human cells infected with bacteria

[Elsie]

Dear all,

I've been asked to help with the bioinformatics for a methylation experiment - human cancer cells infected with two differential bacterial strains from the same species. Within a sample there will be infected and non-infected cells, a methylation profile of the normal human cells and a methylation profile of the infected cells, are there any methylation techniques available that would be able to pick this up? I have alot of experience with HM450 but that just gives a consensus result. What about Methyl-seq?, would it work aligning sequence to a bacterial genome and to a human genome? Any advice greatly appreciated.

[Brian Bushnell]

PacBio reads are supposed to be useful for methylation experiments, which I believe can be run without any special library-prep procedure. I don't know any details or the state of the related software since we don't do methylation studies here, but it's worth looking into, as I have heard it's the best for that purpose.

[nucacidhunter]

Methyl-Seq (WGBS) is the best method to give resolution in single base level and there are kits that enables bisulfite converted library prep from low input amounts (below 50 ng). My concern is mapping converted libraries from cancer cells to reference genome considering complex re-arrangement observed in this type of cells. Other issue to consider is sorting bacterial converted library reads. You may have to consider sequencing bacterial genome as well if there is no reference.

[Brian Bushnell]

FYI, PacBio's methylation detection is supposed to work on all four bases, not just one.

[Elsie]

Thank you Brian and nucacidhunter for your replies.
Brian - I agree, I think PacBio would be a great option to try, however there isn't a machine currently available in Australia and outsourcing is likely to take too long.
nucacidhunter - thanks for comments about sequencing the bacterial genome, I am not sure of the species, but I will check.

[nucacidhunter]

FYI, PacBio's methylation detection is supposed to work on all four bases, not just one.

I totally agree with you. Maybe I should have mentioned for reasonable cost. That is going to cost a lot to do methylome sequencing on PacBio for resonable human genome coverage ( I assume the aim is to know the methylation status in human cells).

[Elsie]

Yes - how does the methylation profile change upon infection.

[dpryan]

I would expect any MethylC-seq aligner could handle your situation. The bacterial genome will be different enough that, even after conversion, it's unlikely to map much/at all to the converted human genome. Further, the rearrangements in the cell line will be there for both groups and you'll only get messed up alignments at the rearrangement borders (which, again, should be similar in both groups, so I wouldn't worry too much about bias then).

If you're worried about the bacterial DNA screwing up alignments, then just try simulating some reads (Sherman would work for that, assuming you have a bacterial reference genome (and if not, bacteria are small, so that shouldn't be difficult to generate)) and see how big of an issue that'd be.

[Elsie]

Thank you dpryan, those are helpful comments.