Bisulfite sequencing bias

[cnicolet]

Hello all:

We have been having a bias problem on our whole genome bisulfite libraries, very severe dropouts in CpG islands. Coverage across the rest of the genome is very good. We are using the Kapa + U enzyme, and a longer denaturation protocol (5' 98° then 4-8 cycles of 40" 98°--30" 65°--1' 72°). But we don't even know if it's a GC issue per se since the CpG islands should have a lot of converted C's due to hypomethylation at these sites. Any insight would be hugely appreciated!

[simonandrews]

What is the overall methylation level you're seeing from your samples? We had one instance of a library which was treated very harshly with bisulphite (stock solution was incorrectly diluted) and we found that the methylated parts were selected for, so we ended up with patchy coverage and a much higher than expected methylation level. This would fit with your CpG islands disappearing since they'd generally be unmethylated.

[C.R.]

Hi, thats an interesting enzyme. It looks like that the 'improved representation of AT-rich sequences' introduces another bias. Maybe you should compare your data with an Agilent Pfu Turbo Cx derived library yourself. On the other hand this enzyme was already tested in many labs, if you look at the KAPA web page. Hard to say what is going on. I would ask KAPA for support on this issue. At least they should exclude that something similar was not observed in the labs they cite as well.

[cnicolet]

Thanks for the advice! Side by side comparison with Pfu shows no differences, but contacting Kapa tech support is a good idea. Methylation levels among the different libraries have not been assessed as a possible variable, thanks for tip.