Help with low input RRBS not aligning

[aleichter]

Hello!
I am trying to optimise a protocol for low cell number input for RRBS library prep. I am having an ongoing issue where the library prep appears to be successful, when checking the FASTQC results there are high quality read scores, good traces indicating MSPI digestion and bisulfite conversion worked and overall the signatures look very similar to our previous bulk (larger input) RRBS libraries which have been successful. However when the libraries are aligned we are consistently only getting 1-5% alignment (even in our larger input libraries 500-1000 cells), these are human cells by the way. We have not identified any contaminant through a blast search and I have now individually tested and combined each of the processes (e.g lysis, MSPI digestion, adapter ligation, and bisulfite conversion) which all appear to be working effectively. I've altered the adapter dilution, and trialled different polymerases, all these conditions appear to have no effect on improving the final output. Has anyone had any similar issues? Or any advice as to what to tackle next?
I have attached the protocol paper which I am following for reference.

[dpryan]

The first question to ask is whether this is really a problem with the library or instead one of the aligner or its settings. What are you using there?

[aleichter]

We suspect it is a library problem instead of the aligner, we use Bismark. In our last batch we ran a positive control with the prepared samples, this was a bulk RRBS library we had previously sequenced with good results, (it was prepared with the protocol we use in our lab for larger input) the alignment for this samples was ~65% which is what we expected. Also worth mentioning, yesterday we performed a hard trim (down to 28 bp I believe or the minimum seed content) our alignment in our samples increased quite a bit up ~20%, this is why I suspect a library issue rather than bioinformatic issue. Do you agree?

To me it seems perhaps it is introduced in the final amplification step seeing as the DNA seems relatively intact (at least up to 200 bp) as demonstrated in some of my PCR tests where I amplified bisulfite specific primers which we use in our lab to check the bisulfite conversion was successful.

[fkrueger]

Hi aleichter,

I'd be happy to take at one of the samples to see if I can find any of the typical suspects in the data. Emailing of a subset of say 500K sequences (gzipped) is typically sufficient for this ([email protected]).

Cheers, Felix