GBS vs RAD

[Jayakumar S]

Hello,
We are planning to study the diversity of a species and for the same we are in a confusion whether to go for RAD or GBS?.
In which technique, we will get more SNPs genotyped?.
What will be the depth in each?.
Kindly suggest.

Thanks
Jayakumar S

[nucacidhunter]

Number of SNPs that can be genotyped is not dependent on the techniques. To detect SNPs firstly they must be present in genomes and that will be dependent on divergence of samples. Secondly, SNP detection is an statistical process and there is always some uncertainty about its existence or detection. With more stringent criteria number of detected SNPs decreases.

RAD or GBS (and many of their derivatives) essentially are reducing representation of genomic regions by their distance from one or two restriction recognition site(s). Depth or coverage will be influenced by two factors. One is proportion of genome that is represented in library (tag number) and second is number of reads obtained for each sample. Generally, a method such as ddRAD that involves a size selection step would be desirable to tune number of tags and sequencing requirements to obtain a suitable depth.

[Jayakumar S]

Hi,
I understand that the number of SNPs that can be genotyped is not dependent on the techniques.
For the specified number of SNPs identified at 20X, in which technique, we will get more number of SNPs genotyped?.
Please correct, if I am wrong.

Thanks
Jayakumar S

[nucacidhunter]

Generally %5 of RAD-tags are polymorphic, so to obtain 1k SNP one needs 20k tags common to all samples if missing genotypes are undesirable. Number of reads/sample will be: number of tags x depth x2. Tag number resulting from any restriction enzyme alone or in combination can be estimated by in silico analysis of species or a close relative genome. For specis with no reference genome it has to be estimated emperically.

[Jayakumar S]

Thank you for the information.
Can you please suggest whether ddRAD or GBS will be better to study the diversity in livestock?.

Thanks
Jayakumar S

[nucacidhunter]

I would suggest searching literature for GBS on the animal that you are going to study and that may give some ideas for your project. If you are going to outsource library prep you may get specific advice for your project from service providers. If you want to prepare libraries in house, I would suggest ddRAD with 90 bp window for size-selection and multiplexing 96 samples for one HiSeq lane. After sequencing and data analysis to increase coverage (if it is lower than desired) you can sequence the same library gain. As I have mentioned before in silico analysis of a related reference genome also would be useful.