ddRADseq library variability

[Shann_rk]

Hi everyone,
I'm in the process of running through the ddRADseq protocol with my samples and I've recently received the first lot of sequences back and it seems something has gone wrong in sample prep.
What we have done is pool ~500 samples (in groups of 48) using a barcode-index system, and run them on the illumina HiSeq platform.
I'm using the STACKs pipeline for most of my analysis (although I may move from this further down the track and look at developing my own... we'll see how much motivation I have) and after running ustacks, it appears that I have major differences in read depth and number of stacks between groups of 48.
Each group was prepared at a different time, but using the same procedure...
So, when I graph number of stacks against number of reads, I get a similar curve between groups, but the number of stacks varies by a lot (so, I've got several curves at different levels if groups are graphed together, if that makes sense). I would maybe expect this between groups if they were different species, but within species I figured I'd get some consistency.
My guess is that its due to 1 of 4 reasons:
1 - Inconsistent size selection
2 - Incomplete/inconsistent digestion
3 - PCR bias
4 - Problem with STACKs
Has anyone encountered this before? And if so, what was the problem? I'd like to start my second lot of library prep, but need to sort this out before I start.

Any advice would be good!
Cheers,
Shannon

[SNPsaurus]

How are you doing the size selection? Even a few basepairs difference in the size range can mean adding or losing hundreds to thousands of loci with ddRAD. Size selection is usually the primary issue with ddRAD. Since you are working within a species the loss and gain of the frequent-cutting restriction site should be less of an issue.

DNA quality can also play a role. Lower-quality DNA can produce more "off-target" loci from improper adapter ligation to free ends of DNA. Can you tell from STACKs if there are lots of loci with a low level of reads?

[nucacidhunter]

Hi Shann_rk
I am wondering if you could post a Bioanalyser trace or gel photo of you single and double digested DNA for a few samples that has made a good library.