I have performed a A11 bpp analysis with a dataset composed of 2 mitochondrial (concatenated) and 2 nuclear genes, all from a supposed one lizard species. I used a Bayesian mithocondrial tree as a starting tree, which suggest that I am dealing with 13 lineages. Previous analysis show deep genetic structure between lineages, which are distant ~10-15% on mithocondrial genes.
The .ctl file, the output from the run, locusrate and scalars are available on the link:
I suspect that the rjMCMC suffered from poor mixing, since there are several possible species that wasn´t even considered by the algorithm. so it did not appeared on the output (e.g. HI). Is that right? I was wondering what could be the best option to get rid of this: maybe I can run the chain longer, or also try to split the data on the two monophyletic groups indicated by the mithocondrial tree (ABCDEFG and HIJLNM), perform two runs, and then a third run with the species delimitated on first runs.
It´s my first time dealing with BPP, so I guess I´m interpreting the results right.
Any suggestion will be greatly appreciated, thanks in advance!!
Last edit: Anonymous 2016-03-26
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
i assume you have gone through the tutorial and read the documentation.
i suggest that you start with the simple. Analyse the mt locus as a dataset, and the 2 nuc loci as a separate dataset. Turn off locusrate and heredity scalars. Those two options are fairly complicated.
ziheng
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thank you for your feedback. As you suggested, I analysed the mt and nuc datasets separated without locusrate and heredity scalars. First I perform a A00 analysis for both datasets to estimate theta and tau priors, as recommended on the tutorial and documentation. After adjusting the priors, I ran the A11 analysis. I already have the results from nuc dataset, which you can download on the following link (as well as the ctl file):
The mt analysis is still running, but I believe that, at least for the results above, the chains are still suffering of poor mixing. What is your opinion? Do you (or others) have any suggestions to get rid of this? I was thinking in run the chain longer, and also use speciesmodelprior=3, that, according to the manual, may be suitable when there is a large number of populations (13 in my case).
Best regards
Sergio
Last edit: Anonymous 2016-03-31
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
View and moderate all "BP&P Q&A" comments posted by this user
Mark all as spam, and block user from posting to "Discussion"
Hello,
I have performed a A11 bpp analysis with a dataset composed of 2 mitochondrial (concatenated) and 2 nuclear genes, all from a supposed one lizard species. I used a Bayesian mithocondrial tree as a starting tree, which suggest that I am dealing with 13 lineages. Previous analysis show deep genetic structure between lineages, which are distant ~10-15% on mithocondrial genes.
The .ctl file, the output from the run, locusrate and scalars are available on the link:
https://drive.google.com/folderview?id=0B7NeSA7GlnZZa2ZTVElYOEd3M0U&usp=sharing
I suspect that the rjMCMC suffered from poor mixing, since there are several possible species that wasn´t even considered by the algorithm. so it did not appeared on the output (e.g. HI). Is that right? I was wondering what could be the best option to get rid of this: maybe I can run the chain longer, or also try to split the data on the two monophyletic groups indicated by the mithocondrial tree (ABCDEFG and HIJLNM), perform two runs, and then a third run with the species delimitated on first runs.
It´s my first time dealing with BPP, so I guess I´m interpreting the results right.
Any suggestion will be greatly appreciated, thanks in advance!!
Last edit: Anonymous 2016-03-26
i assume you have gone through the tutorial and read the documentation.
i suggest that you start with the simple. Analyse the mt locus as a dataset, and the 2 nuc loci as a separate dataset. Turn off locusrate and heredity scalars. Those two options are fairly complicated.
ziheng
View and moderate all "BP&P Q&A" comments posted by this user
Mark all as spam, and block user from posting to "Discussion"
Dear Dr. Ziheng,
Thank you for your feedback. As you suggested, I analysed the mt and nuc datasets separated without locusrate and heredity scalars. First I perform a A00 analysis for both datasets to estimate theta and tau priors, as recommended on the tutorial and documentation. After adjusting the priors, I ran the A11 analysis. I already have the results from nuc dataset, which you can download on the following link (as well as the ctl file):
https://drive.google.com/folderview?id=0B7NeSA7GlnZZaDJvTTdtQXhTMGc&usp=sharing
The mt analysis is still running, but I believe that, at least for the results above, the chains are still suffering of poor mixing. What is your opinion? Do you (or others) have any suggestions to get rid of this? I was thinking in run the chain longer, and also use speciesmodelprior=3, that, according to the manual, may be suitable when there is a large number of populations (13 in my case).
Best regards
Sergio
Last edit: Anonymous 2016-03-31
View and moderate all "BP&P Q&A" comments posted by this user
Mark all as spam, and block user from posting to "Discussion"
Hi Sergio,
I'm suffering from similar problems, could you get rid of them? If so, could you please tell me how? Thank you!
Best,
Carlos