I'm planning to do a population genetic analyses (genetic differentiation, diversity) on a bee species using SNP markers. Furthermore, I aim to identify genes under selection. For population genetic studies based on microsatellites, ususally it is recommended to sample at least 30 individuals per population to have enough power. Does this recommendation also hold for SNPs, since they are biallelic and thus less variable than microsatellites (but also more abundant)? What is your opinion about this? How many samples do I need for the population genetic part and how many for the study of adaptation?
Recently I came across a similar question for a fish species. Remember browsing through a few articles. Some information that I could quickly browse again for you:
"Given that each SNP marker has an individual evolutionary history, we calculated that the most complete and unbiased representation of the genetic diversity present in the individual can be obtained by incorporating at least 10 individuals into the discovery sample set, to ensure the discovery of both common and rare polymorphisms" [PMID: 21207104 - a study on primates]
Other potentially important articles (pubmed IDs):
22685040
18288444
20555335
17227482
16258542
16137993
15812170
11836209
Belinda, you partially answered your question. There is not a "magic number" for that. I think that 30 is a number sufficient to obtain statistical reliability,
I agree with Ferrucio, there is no "border number" (like it is for PCA analyses, I found 100 observation is lower). However another task is how many you need if plan to publish it in some journal or present on conference (just don't know what are demands for non-human analyses).
Cohen's d and Hedge's g are to be used to compute sample size effect (available in statistical packages and R language). For a smaller sample, Hedge's g is better to be used, after performing statistical analyses (ANOVA for example if data is parametric).
Belinda, it depends what type of population and what for you want to study...For instant, in genetic isolates for study genomc polymorphism using any DNA markers you will have more work with selection of unrelated subjects (randomized samples-one of requirement of gene pool study in population)...regarding representative sampling -in genetic isolates you may collect samples 4 times less than in outbred-genetically heterogeneous population -to characterizes it's gene pool.
-In reality there is one 'old-fashion' criterias: In my young ages -we -groups of young population-geneticists used immunological loci ABO, MN,Rh and some biochemical markers and almost every day after collecting samples in our expeditions analysed polymorphism of the alleles,,,when after several days of intensive collections and lab analyses result we used for polymorphism calculations, if we found that during 3 following days our results on alleles polymorphisms are not changed we knew that already collected data are enogh for gene pool characterization in population under study.
Thanks Kevin, I downloaded the article as well...But in any case no one software can define what type of population you are going to study -this is a researcher responsibility. Any population-genetic processes will depend from type of populations under study, for instant, genetic bottleneck will different in genetic isolates, than in outbred populations.
Thank you, Kevin, I downloaded your interesting article as well. it's known that .sizes of samples in populations genetics study depends from Ntotal of the population under study, as well as from it's type. And another criteria that showed founders of population-genetics in mathematical modelling: with smaller samples characterized by numerous traits (i.e. 100) we can get same information as in larger samples /groups of subjects (i..e. 100) with limited traits....sorry my english:-) In my own studiees I prefer to use classical criterias - to organized a randomized and representative group in every population under study.
Thanks a lot for your help and the papers. Unfortunately there is not much known about my study species and it is not possible to genetically analyse samples directly after sampling. Thus I will not know if after lets say 7 days of sampling I already have collected enough representative individuals, by measuring allele freq cange as suggested. However, with the help of your papers and answers I got a much better idea for my sampling design.
Recently I came across a similar question for a fish species. Remember browsing through a few articles. Some information that I could quickly browse again for you:
"Given that each SNP marker has an individual evolutionary history, we calculated that the most complete and unbiased representation of the genetic diversity present in the individual can be obtained by incorporating at least 10 individuals into the discovery sample set, to ensure the discovery of both common and rare polymorphisms" [PMID: 21207104 - a study on primates]
Other potentially important articles (pubmed IDs):
22685040
18288444
20555335
17227482
16258542
16137993
15812170
11836209
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size