popSize value

[jssd...@gmail.com]

Sorry for such a simple question, but I was not sure about the units for popSize in the coalescent population size models, based on the explanation in the Parameters section of the BEAST2 website.

If I have an estimate of 2 for popSize, and have calibrated the root using an outgroup with divergence in millions of years (e.g. root age 7), and popSize = N*t, am I dividing by 1 (units of geological time), 1e6 (millions of years), or some other value, to get N_e? It seems odd that the estimate for N_e would be single digits or smaller.

[Tim Vaughan]

It's a good question.  The only not-wrong thing you can really say about the "popSize" estimate that BEAST is giving you is that it's the inverse pairwise coalescent rate.  If you're happy to assume that your data were generated under a Wright-Fisher model then additionally you can say that this paramter popSize = Ne*t where t is the time between _generations_, whatever that means in your context.  So in your case if your generation time was ~1y you'd need to divide the popSize parameter by 1e-6 in order to get an estimate of Ne.  (Of course ploidy also needs to be taken into account if you're serious about getting the true Ne, so additional factors may be involved.)

Hope this helps,

Tim

On Thu, 14 Mar 2019 at 22:05, <jssd...@gmail.com> wrote:

Sorry for such a simple question, but I was not sure about the units for popSize in the coalescent population size models, based on the explanation in the Parameters section of the BEAST2 website.

If I have an estimate of 2 for popSize, and have calibrated the root using an outgroup with divergence in millions of years (e.g. root age 7), and popSize = N*t, am I dividing by 1 (units of geological time), 1e6 (millions of years), or some other value, to get N_e? It seems odd that the estimate for N_e would be single digits or smaller.

--
You received this message because you are subscribed to the Google Groups "beast-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to beast-users...@googlegroups.com.
To post to this group, send email to beast...@googlegroups.com.
Visit this group at https://groups.google.com/group/beast-users.
For more options, visit https://groups.google.com/d/optout.

[jssd...@gmail.com]

Thanks, Tim. I had been exploring the estimate on a haploid dataset, and was getting results for popSize between 2 and 6. We aren't really sure of the generation time, but it may be on the order of a decade or so.

[Artem B]

Hello!

The Pop.Size parameter is Θ = Ne*τ, where τ is generation time in the units of a molecular clock calibrated by you. I think it is reasonably to consider the τ as the real generation time of a studied organism. For example, If the organism has generation time about 20 years and your clock calibrated as My, then τ is 0,00002. So the Pop.Size you get divided by 0,00002 to get Ne. All you need to know is true generation time of the studied haploid organism. In the case of an infection agent under an investigation, generation time is a time interval between two transmissions.

Regards,
Artem

[mumm...@gmail.com]

Dear sir:

     I am still  not quite understand the meaning of the parameter PopulationSize in the Multi-type  Model in BEAST2.5.2. How should i choose the value of it. In different model the  PopulationSize has different default values.

   Thanks

Jinjin
在 2019年3月15日星期五 UTC+8下午4:17:02，Tim Vaughan写道：

It's a good question.  The only not-wrong thing you can really say about the "popSize" estimate that BEAST is giving you is that it's the inverse pairwise coalescent rate.  If you're happy to assume that your data were generated under a Wright-Fisher model then additionally you can say that this paramter popSize = Ne*t where t is the time between _generations_, whatever that means in your context.  So in your case if your generation time was ~1y you'd need to divide the popSize parameter by 1e-6 in order to get an estimate of Ne.  (Of course ploidy also needs to be taken into account if you're serious about getting the true Ne, so additional factors may be involved.)

Hope this helps,

Tim

On Thu, 14 Mar 2019 at 22:05, <jssd...@gmail.com> wrote:

Sorry for such a simple question, but I was not sure about the units for popSize in the coalescent population size models, based on the explanation in the Parameters section of the BEAST2 website.

If I have an estimate of 2 for popSize, and have calibrated the root using an outgroup with divergence in millions of years (e.g. root age 7), and popSize = N*t, am I dividing by 1 (units of geological time), 1e6 (millions of years), or some other value, to get N_e? It seems odd that the estimate for N_e would be single digits or smaller.

--
You received this message because you are subscribed to the Google Groups "beast-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to beast...@googlegroups.com.

[Tim Vaughan]

Hi Jinjin,

It may have different default starting values in the MultiTypeTree BEAUti interface, but it still means the same thing - just generalized to more than one sub-population according to the structured coalescent.  The starting value itself isn't tremendously important (shouldn't affect your results provided you run until convergence) but the prior on the parameter certainly does, and for the sub-population size you can apply the same or similar considerations to the ones you use when setting up the prior for the unstructured model.

Tim

To unsubscribe from this group and stop receiving emails from it, send an email to beast-users...@googlegroups.com.

[mumm...@gmail.com]

Hi Tim：

     Thanks. Another question about MultiTypeTree BEAUti interface,I use the HIV pol sequences labeled by different city and different sampling time and i am not sure the sampling time from different cities should be the same.For example,if i have 5 sequences sampled at 2008 to 2012 in  London and so i must have other 5 sequences also sampled from 2008 to 2012 in America? In the MultiTypeTree tutorial，i found the the number of sequences from different city is the same,so i guess we should keep the same,but i do not know about the sampling time.

     Best
Jinjin
在 2019年4月15日星期一 UTC+8下午3:30:37，Tim Vaughan写道：

[Tim Vaughan]

Hi Jinjin,

In general the SC model places no hard constraints on when and where the samples are from.  I.e. you'll get a sensible posterior for your migration rates, pop sizes and tree in all cases.  However, analyses based on a good number of sequences from each location and a range of times generally produce the most informative results.

Tim

To unsubscribe from this group and stop receiving emails from it, send an email to beast-users...@googlegroups.com.

[HS]

Dear Tim,

I would like to run your method with human Y chromosomal data In search of the place for the root origin. Does it make sense, or MTT is for only viruses and  serially sampled data?

Best,

Hovhannes

[Tim Vaughan]

Dear Hovhannes,

MTT is a direct implementation of the basic constant population size structured coalescent model, so wherever that model is appropriate you can use MTT.  There's no reason that you cannot use it to analyze Y chromosome data, although the assumption of constant population size might be a bit ridiculous, depending on the sample set.

Tim

--

[mumm...@gmail.com]

Hi Tim:

    Thanks.

     I run BEAST recently always the same error'Another program is using this file process unreachable'.Even stranger, this error occurs after running 2 million MCMC chain length. I also closed or changed the related .xml files names before run BEAST.(It's the same error with my other teammates) I have downloaded two versions BEAST V2.4.8 and V2.5.2,now i need use V2.5.2,should i only keep one version BEAST in my computer?

    Best.

Jinjin

在 2019年4月16日星期二 UTC+8下午3:24:52，Tim Vaughan写道：

[Carlo Pacioni]

Hi Jinjin,

This often happens when beast is trying to write the state file. Try to inrease the interval by 10 folds. That should fix it.

Carlo

To unsubscribe from this group and stop receiving emails from it, send an email to beast-users...@googlegroups.com.

[HS]

Dear Tim,

Thank you very much for your reply!

I got the results now. I have 111 sequences, and Bayesian Skyline analysis show intensive population growth as many Y lineages. I split my dataset into 2 populstions - Northern (n=33) and Southern (n=78). I get plausible popsize estimate for the Northern one, but for the southern population the size stayed uniform as was the prior. Can't understand why there is no info to estimate the size for this population as well.

Another issue I don't understand is the assignment of the branch that is completely from the Southern sequences to the North.

Do you have any ideas what I can try more?

Best,

Hovhannes

On Tuesday, 23 April 2019 18:30:38 UTC+3, Tim Vaughan wrote:

Dear Hovhannes,

MTT is a direct implementation of the basic constant population size structured coalescent model, so wherever that model is appropriate you can use MTT.  There's no reason that you cannot use it to analyze Y chromosome data, although the assumption of constant population size might be a bit ridiculous, depending on the sample set.

Tim

On Tue, 16 Apr 2019 at 16:09, HS <hovo...@gmail.com> wrote:

Dear Tim,

I would like to run your method with human Y chromosomal data In search of the place for the root origin. Does it make sense, or MTT is for only viruses and  serially sampled data?

Best,

Hovhannes

--
You received this message because you are subscribed to the Google Groups "beast-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to beast...@googlegroups.com.