Hi Tonapha,

The tests that you used (Tajima's D, Fu&Li D, Fu&Li F and Fu's Fs) have been proposed to detect departures of nucleotide variability patterns from neutral expectations (the null hypothesis). When departures from neutrality are detected (significant D or F values), very different kinds of evolutionary processes--such as selective events or demographic changes--are invoked to explain these departures. This is the case because both selective events or demographic changes produce similar deviations in these tests. Therefore the interpretation of departures from neutrality is difficult if someone uses only these tests. In such cases coalescent simulations are used to model the impact of a demography on sequence diversity (and recombination).

The fact that you obtained non-significant D and F values from your data suggests that you cannot reject the null hypothesis of neutrality. Therefore the pattern of variability that you observe in this specific gene you are studying among these populations can be explained by the neutral process. If you had molecular data from another gene the case might have been different. To make inferences for a species you need data from multiple genes. And also to incorporate in your discussion information related to the function of the gene, if it is generally conserved or evolves rapidly between species, etc.

On another note, to perform HKA or MK tests you need an outgroup. Do you have such data?

As for calculating dN/dS rates with population data, I would be very skeptical. As I mentioned in one of your previous questions there are issues with the application of the dN/dS ratio with population data. Please see:

http://www.ncbi.nlm.nih.gov/pubmed/19081788

PLoS Genet. 2008 Dec;4(12):e1000304. Epub 2008 Dec 12.

The population genetics of dN/dS.

Kryazhimskiy S, Plotkin JB.

I would also suggest:

http://www.ncbi.nlm.nih.gov/pubmed/18493071

Genetics. 2008 May;179(1):555-67.

Statistical power analysis of neutrality tests under demographic expansions, contractions and bottlenecks with recombination.

Ramírez-Soriano A, Ramos-Onsins SE, Rozas J, Calafell F, Navarro A.

Also you state that you found one codon under positive selection and 5 codons under negative selection. Which tool(s)/ method(s) did you use? What are the assumptions behind the method(s)? Do your data meet the assumptions?

I hope this helps a bit.

Dimitra

When it comes to assessing the selection I would prefer to work on codon substitutions. However, the methods you mentioned are not mutually exlusive so why choose one or another? Unless you have some constrains reagrding the kind of data that you can gather? It would help if you could describe your problem in more details.

Hi Marcin, thank you. My problem is after I tested for neutrality of the sequences of my gene across my populations (one specie but different ecotypes), I found non significant Tajima's D, Fu&Li D, Fu&Li F and Fu's Fs which I guess that these tests are sensitive to demographic history. However, when I analysed the data using dN and dS approach (by estimates of dN significantly different from dS provide convincing evidence for nonneutral evolution) I found one codon under positive selection and 5 codon under negative selection. So, I have an awkward feeling that how should I explain the results of this data?... is this make any sense? .... Thank you.

The non-significant results could be due to low power. Some of the tests are sensitive to demographic history, but usually it leads to significant results rather than non-significant results.

You can go for extended haplotype honozygosity (Sabetti) analysis for your codons... or may go for iHS score's (Pritchard) for recent positive selection check.

Hi Xu and Sarkar, thank you for your suggestions and the literature.

As some of the testes are sensitive to the demographic history. To explain that effect, do demographic modeling analysis needed? such as bottleneck, pop subdivision, or pop expansion.

I would recommend codon models. There's one in particular, MEME, that's been recently published that can identify codons under selection plus lineages or branches that are under selection. Here's the reference:

http://www.plosgenetics.org/article/info%3Adoi%2F10.1371%2Fjournal.pgen.1002764

Thanks Eduardo, I appreciate! I analyzed my seq data using DataMonkey and yes there is MEME. I'm studying and trying to understand about the MEME and the interpretation!

Hi Tonapha,

The tests that you used (Tajima's D, Fu&Li D, Fu&Li F and Fu's Fs) have been proposed to detect departures of nucleotide variability patterns from neutral expectations (the null hypothesis). When departures from neutrality are detected (significant D or F values), very different kinds of evolutionary processes--such as selective events or demographic changes--are invoked to explain these departures. This is the case because both selective events or demographic changes produce similar deviations in these tests. Therefore the interpretation of departures from neutrality is difficult if someone uses only these tests. In such cases coalescent simulations are used to model the impact of a demography on sequence diversity (and recombination).

The fact that you obtained non-significant D and F values from your data suggests that you cannot reject the null hypothesis of neutrality. Therefore the pattern of variability that you observe in this specific gene you are studying among these populations can be explained by the neutral process. If you had molecular data from another gene the case might have been different. To make inferences for a species you need data from multiple genes. And also to incorporate in your discussion information related to the function of the gene, if it is generally conserved or evolves rapidly between species, etc.

On another note, to perform HKA or MK tests you need an outgroup. Do you have such data?

As for calculating dN/dS rates with population data, I would be very skeptical. As I mentioned in one of your previous questions there are issues with the application of the dN/dS ratio with population data. Please see:

http://www.ncbi.nlm.nih.gov/pubmed/19081788

PLoS Genet. 2008 Dec;4(12):e1000304. Epub 2008 Dec 12.

The population genetics of dN/dS.

Kryazhimskiy S, Plotkin JB.

I would also suggest:

http://www.ncbi.nlm.nih.gov/pubmed/18493071

Genetics. 2008 May;179(1):555-67.

Statistical power analysis of neutrality tests under demographic expansions, contractions and bottlenecks with recombination.

Ramírez-Soriano A, Ramos-Onsins SE, Rozas J, Calafell F, Navarro A.

Also you state that you found one codon under positive selection and 5 codons under negative selection. Which tool(s)/ method(s) did you use? What are the assumptions behind the method(s)? Do your data meet the assumptions?

I hope this helps a bit.

Dimitra

Hi Dimitra, Thank you very much again for your very helpful explanation. I now have better understanding. I appreciate.

For the HKA and MK tests, I have the outgroup (sister species - pretty close). Actually, for the Tajima's D there is one of two ecotype has been found significant. That why I performed the dN/dS approach.

For the codon that I have found that under positive and negative selection I used DataMonkey. Do my data meet the assumptions? I'm not quite sure yet. I will try to look at them again.

Tong

Beware that tests that are designed for between species comparisons might not be appropriate within species. I'd also recommend linkage disequilibrium based approaches (extended haplotype homozygosity). Keep in mind that if the two populations haven't diverged for a long time, recombination might not have had enough time to break linkage among selected and non-selected sites.

Interesting point about the tests that may not appropriate for within species!

I'm kind of new for the extended haplotype homozygosity test. I will try that test, any caution?

By the way, according to DNAsp and DataMonkey analysis I found one recombination breakpoint within my gene.

please visit this link http://pritch.bsd.uchicago.edu/software.html... you will automatically get to know about everything. Only one issues, please consult the mother papers, I have already mentioned earlier

Thank you very much!

Hi All, Thank you very much for all your suggestions. I am now understand why dN/dS ratio dose not work well for population study.

I would highly recommend Neilsen's Review for a general background and some suggestions on tests.

http://www.annualreviews.org/doi/pdf/10.1146/annurev.genet.39.073003.112420

Do you mean selective sweep?

In my opinion, I think you should do the LD test between 2 expected to be linked loci to see how those 2 loci link together, highly link or loosely link. Then, think about the allele under selection must be fixed at some point if the direct selection is strong enough. One more think I could think about now is the patterns of neutral diversity at a neutral locus linked to a selected locus can be modeled by conditioning on the trajectory of the frequency of the selected allele through time, and treating the two alleles classes as subpopulations within each of which the dynamics are neutral, with recombination moving lineages between the two.

How do you think? I might be wrong.

Here the paper for reference. http://www.genetics.org/content/early/2012/06/12/genetics.112.141861.full.pdf

We recently wrote quite a big review that covers this topic, it's in a book chapter but you can find it on my website (search for Anisimova and Liberles 2012):

http://www.inf.ethz.ch/personal/anmaria/publications.html

For more details on population genetics tests, I always found Nielsen's reviews very clear and useful,

Thank you very much, Maria!

how can I conclude, when Tajima’s D is positive and insignificant ;and Fu’s Fs is negative and significant?