Given a standard diversity at SNP markers (H around 0.15-0.20), how many SNPs would you need to reliably identify the following sets of relations: parent-offspring, full-sibs, half-sibs, grandparent-grandoffspring, uncle-cousin relations etcetera?
It really depends on how do you define "reliably" for your needs - you need just a probable estimate or a piece of evidence for a lawsuit...? I am not working in this field, but I would imagine the number of needed SNPs in high 10s or low 100s.
Haha, no it's "just" for estimating kinship in ecological research among sampled animals of different species (wild boar, stone marten, toads, newts, ... But also plant species). P
Take a look at this page, it may help you http://www.isogg.org/wiki/Autosomal_DNA_statistics
Aniway you should look at wich fraction of all SNPs within the gene you are taking in the analysis (if they are enough to conclude something about the whole gene or not) and then looking at the percentages in the link I attached to know which kind of kinship are you finding
To estimate kinship relationships without any prior knowledge, we needed close to 150 SNPs to reliably estimate parent-offspring and fullsib relationships. Anything beyond that we could not realiably detect (not even with 192 SNPs). My guess (not based on any calculation or fact) would be that you would need at least 2x192SNPs to detect half-sibs if your minor allele frequency is high. If most of your SNPs have a low minor allele frequency the number most probably needs to be higher. But again this is just a guess from what we saw with our data.
There is one paper in which they estimate how many SNPs are needed. This might help you to find some calculations:
Genetics. Apr 2006; 172(4): 2567–2582.
doi: 10.1534/genetics.105.048074
PMCID: PMC1456362
The Power of Single-Nucleotide Polymorphisms for Large-Scale Parentage Inference
Eric C. Anderson1 and John Carlos Garza
Thanks to all of you for taking the time to answer. it's been very helpful!
- Badges
- Science topic