Arlequin has been developed mainly under WinNT, and it will certainly be the platform under which the fewer bugs will remain. Note however, that the C++ code for the computations and the Java code for the Interface are absolutely identical among plaforms. Thus differences in program behavior will be due to differences in the compilers we have used (Borland C++ 5.02 for Win9X/NT/2000, CodeWarrior for the MacOS, and gcc for Linux) and different impementations of the Java Runtime Environment on different platforms.
The conversion utility provided with Arlequin is not completely bullet-proof against small departure from native formats (e.g. tabs used instead of white spaces...). It implies that the conversion methods may not work if the source format are not exactly as described in user manuals, even if the same files work with other packages.
We may try to develop a version for Linux on the Mac, and perhaps for Solaris on Sun. If the present code run without too much problems we'll do it, otherwise we fear not to have enough time for fine-tuning our code for these platforms.
If some people are ready to test our code on other platforms and make it compile, then they should contact us.
Schneider, S.,Roessli, D., and Excoffier, L. (2000) Arlequin: A software for population genetics data analysis. Ver 2.000. Genetics and Biometry Lab, Dept. of Anthropology, University of Geneva.
Negative variance components can sometimes occur, because they are rather covariances. Their associated fixation indices can also be seen as correlation coefficients. Usually, slightly negative variance components can occur in absence of genetic structure, because the true value of the parameter you want to estimate is zero. Thus, if the expectation of the estimator is zero, you can have, by chance, slightly positive or slightly negative variance components. Most of the time, these negative variance components indicate an absence of genetic structure. They can have a biological meaning, though. For instance, in outcrossing organisms, genes from different populations can be more related to each other than genes from the same population.
It really depends which kind of estimator you want to use for AMOVA. If you want to compute an anlogue of Slatkin's RST, then you would need to provide your data coded in terms of absolute or relative number of microsatellite motif repetitions. This is needed because we need to compute the sum of the square number of repeat differences between each pair of microsatellite haplotype. If your input is proportional to the mere length of the amplified PCR product, this estimation will be flawed. In this case, we would advise you to rather compute a FST-type statistic, which do not use information on the amount of difference between alleles at each microsatelite locus.
Yes, you can use the AMOVA procedure to test for the presence of genetic structure under certain conditions. However, you must assume that you have the same mating pattern in all your population samples. If you are ready to make this assumption, you can pretend you have RFLP markers, and proceed with the AMOVA analysis. In that case you are going to partition the genotypic variance, and not the variance of allele frequencies as for co-dominant markers. Therefore, even though the proportion of variance due to different levels will be quite informative,and its significance will be meaningful, you should not try to compare the estimated F-statistics to those inferred from co-dominant markers. This is because F-statistics refer to correlation of genes.for dominant markers, but they would be here equal to correlations of genotypes.
We will incorporate it when we'll need it for our personal purpose. Sorry guys, but we only have one life.
This is because you did not configure Arlequin properly. See p. 45 of the manual.
You need to go the Configuration tab and press the Browse buttons to select which applications you want for browsing the result files (now in HTML format) and for editing your project.
This may be due to one of the following problems:
FIS estimates can only be computed by the conventional AMOVA procedure if the gametic phase is known. So if the phase is known then you should be able to check the checkbox "Include individual level for genotypic data" in the "Genetic Structure | AMOVA" dialog box.
If the gametic phase is not known, then you can use a trick:
Arlequin output MSTs in the NEXUS format that can be visualized using, for instance, the free program TreeView created by Rod Page.
Last update : 25.10.00 (08:32)