Hi all,
In some of my work, I am dealing with degraded DNA extracted from a sediment core that did not have the best sedimentation conditions. We know that the DNA is degraded, with a mean size of 250bp. We would like to quantify the frequency of a few SNP at sites that we already have found in non degraded DNA through RADseq.
I know RADseq has been done on degraded DNA, with modified protocols: no sonication is necessary since fragments are already small. However, a second modification to the original protocol makes me raise some questions:
Degraded DNA is fragmented, but also have a lot of single stranded DNA (ssDNA). Restriction enzyme used for RADseq (here sbfI) will not cut if the site is single stranded, and fragments. The modification of the protocol is then to use the Klenow fragment to synthesise double stranded DNA (dsDNA) from ssDNA (in our case,denatured dsDNA and ssDNA from degradation) and thus try to basically "fix" ssDNA areas before using RE digestion, in order to have the maximum number of cut sites available to us.
Although this is perfect in order to recover as much information as possible, the fact that ssDNA degrades faster than dsDNA might be a problem for SNP calls, and many reports show an increased error in nucleotide call near the break points of aDNA, because of chemical changes (like deamination) in the DNA molecule and thus an impossibility for the polymerase to detect the proper nucleotide.
My question is thus: is it worth doing the synthesis of the complementary strand of DNA prior to do RE digestion, even if some RADtags might contain extensive amount of nucleotide substitution, or should I be conservative and go with RE digestion first, hence loosing any restriction sites in ssDNA regions but ascertaining the quality of the nucleotide sequence of each cut site?
In a way, it is like asking: if I do complementary synthesis before digestion, will the fragments I recover be trustworthy and generally pass strict mapping filters...or am I loosing precious sequencing resources to get data on cut sites that I won't be able to use at the end?
Celine, I think that you may want to ask Beth Shapiro or Ed Green, at UCSC, about this, as they encounter this kind of problem with their ancient DNAs. I don't know if they are on RG so you may want to email them directly. Cheers. Giacomo
Have you considered creating pulldown baits using the original RAD data and then capturing the ancient DNA?
It would be good to test what proportion of the DNA is dsDNA. If it is a very small fraction, you might need to do so many PCR cycles the sequencing library would be full of errors anyway.
Sounds like a challenging problem!
Hi Giacomo! Good to see you here! Beth Shapiro would seem the best person to ask indeed. I will send her an email.
Eric, yes we actually thought about it, we wanted to create a bait reaction (MYbaits target enrichment kit) and get only the sites that interested us, however the cost of designing the bait is huge for us, and as you mentionned, several round of PCR might be necessary to get enough data to sequence. Also, we have an aDNA plateform here in Montpellier, but people are mostly doing shotgun sequencing, sometimes with a whole genome duplication prior to sequencing to insure that they have enough material. For us however, since the goal is a potential estimation of allelic frequencies in a pooled sample (quantity of DNA should not be too much of an issue), we need depth, and shotgun will favor genome coverage versus depth.
I think combining quantification methods for dsDNA and BioAnalyzer profiles of 1. Extracted DNA, 2. Digested DNA without second strand synthesis, and 3. Digested DNA with a prior second strand synthesis. This will give us an idea of what our library could contain in term of size class and quantity, and to see if synthesizing the second strand prior to digestion really helps recovering information...
It is indeed challenging!
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology