Hello there.
What I mean: I have sequences of one mitochondrial gene of one species from different regions of the world. I see micro-variation in it (haplotypes). It's difficult to get a specific and good value to describe diversity of the networks I get with my data. So I thought about using a molecular clock to compare my data. At best I want to get an estimated time-value of evolution for each of my datasets. So that I could say for example: In place A there we have 1.2 million estimated years passed and in comparison to that in place B there have only 0.92 million estimated years passed.
Is there a software to use for such a thing? I searched already, but only found clocks for interspecific questions, or to get a timeline for some trees and so on.
Thank you all in advance for your help.
BEAST has been used for these sort of things. See papers by beth Shapiro and Simon Ho.
There is also extensive coalescent literature in population genetics, especially in the context of studies on human evolution. It is mostly driven by genomic data though.
BEAST is commonly used for this type of work. There is extensive discussion of how to use it for such applications in the manual.
If your regions are monophyletic then it's a relatively simple problem to date a node as long as you have a reasonable estimate of the rate - Weston and Krzystof's suggestions to use BEAST are good ones. But if you've got haplotype sharing or incomplete sorting it gets more complicated. I'd suggest using IM or IMa which uses Hey and Neilsen (2007) coalescent method to estimate population divergence with migration. It's been around for a while so there's quite a bit of literature on it. You can get the software here:
https://bio.cst.temple.edu/~hey/software/software.htm
Using coalescent samplers such as BEAST and IMa is important as noted above. You also need to be careful of "time-dependency" in which rates calibrated at intraspecific time scales (say 1 million years or so) are *apparently* faster than rates calibrated from older time points (e.g. fossils or the Isthmus of Panama). Hypothesized reasons for this phenomenon have to do with nearly neutral theory and weak purifying selection. So if you use a (slow) interspecific rate on an intraspecific estimate, you will likely get dates that are biased high. You might search the literature for ancient DNA calibrations on something related to your taxon, or alternatively I recently wrote a paper that allows one to make relatively recent calibrations from known expansion events (attached) :-)
Article Time Dependency of Molecular Rate Estimates and Systematic O...
Article Expansion Dating: Calibrating Molecular Clocks in Marine Spe...
Thank you very much to all of you.
Now I have new things to work on. ;-)
Eric's answer is very good. The molecular clock calibration (a mutation rate) in IMa can be a point estimate but it is also possible to give the algorithm a range of possible mutation rates that reflect uncertainty about how good the molecular clock calibration is (for the reasons Eric noted). Using a range rather than a single rate causes the model to converge on a broader range of likely parameter values (especially the divergence time parameter). Good way to avoid overconfidence in parameter estimates.
Thank you very much for all your answers. I have some problems now. I'm using samples of Echinococcus granulosus. As you may imagine, you can't have samples all the time you want them, so I have a fixed data set. So it is impossible for me to calculate a mutation rate from this samples to calibrate the clock. And I don't know any fixed occurence in the past of E. granulosus to use as calibration point. What can I do now?
And a second question: Is there a instruction or explanation how to use a MC or what it does exactly, which is a little bit easier to understand? Like: "For Dummies". I'm completely new in this topic, so its really hard for me to understand everything about the parameters, Markov Chains and so on.
You could use the interspecific clock calibrations, and adjust upward to include a range of faster rates (as Artur and Eric suggested).
No, there is no 'Dummies' instruction for MCMCMC methods. It's really important to understand the details of how the models work, and how to get good model fits (and good parameter value estimates). There are some really bad IMa analyses in the published literature by users who didn't understand how the software and the models work. One way around this is to find an experienced IMa user who wants to collaborate with you, and can teach you how to use the models.
Not sure about IM, but BEAST comes with very good tutorials that walk you step-by-step and explain how to build a model from very simple to very complicated, and how to test its validity. Remco Bouckaert wrote a free book on how to use BEAST2 (http://beast2.org/book.html).
- Badges
- Science topic
- Similar topics
- Biological Science
- Biota
- Eukaryota
- Fungi