I have targeted methylation sequencing (Roche Nimblegen kit) from FFPE tissue. I know there are several issues with FFPE, but I am seeing a very specific result of using this material. The bias plots all start off with low GC in the first 40-50 bp, then increase to the 'normal' level of around 50% (see plot). I have used bwameth for alignment and bias plotting (this work was done a while ago but is only now being used for interpretation/papers; I see bwameth authors are specifying to use PileOMeth to tabulate and plot now).
Any ideas on what this effect is would be of great help, I thought possibly adapter contamination, but have trimmed adapters, and also believe Illumina adapters are methylated, so shouldn't reduce the proportions as seen(?)
Yikes, how big are the targeted regions? If they're big enough that you have coverage from a lot of different positions within the reads then this is salvageable. Otherwise you'll end up finding a LOT of false positives (or noise, depending on what you're doing) due to this.
Targeted region histogram added, majority are 100bp, total 86MB.
My rationale was, as you say, that if there are enough reads covering the targets then there should be an even distribution of reads over targets and the low CG at 0-40 bp shouldn't matter. Data is noisy for sure.
But my main issue is: do you have any idea what is causing this? I am totally perplexed!
I have no clue why that's actually happening :(
It's probably the GC clamp no? Particularly if you have asked for primers that capture whole genes (where surrounding DNA is generally AT-rich, and genes are GC rich).
EDIT: thought you meant Illumina adapters.
OK, this makes sense, we did include TSS regions so will check with Roche and get back about it. Thanks for the idea.
Not the adaptors, the primers used for the targeted sequencing. They typically end (3') with a high GC-content, which would of course on a plot like this make everything 5' look like low-GC.
Ah, that makes sense.
Very nice, I will look into it and revert when we get word back from Roche.
Hi,I wonder how to draw a mbias plot using bwameth ? I cant find the script on github.Thank you very much!