Hi,
I have tried BEAST for calculating the evolutionary divergence time for my study, consisting of ~ 80 taxa & having a length of ~80,000 bps per taxa.
After the run was completed, the BEAST gave the output which I couldn't interpret. In the acceptance rate column, for various operators, it gave the suggestion to: Set the scale factor to some values..
How to set the scale factor? Also the posterior ESS & likelihood values became infinity after a few thousand sample runs.
In the tracer output also, the ESS values are all very small. Also, the mean, stdev, variance and all other values are also not shown properly.
I need to calculate the divergence time of a particular gene and am struck with this. Any advice on this would be highly helpful.
Hi Sunitha,
Have you tried to do the tutorial first? I have attached one of them for you, but there are many more out there for different versions of BEAST.
Best of luck,
Luther
Hi Sunitha,
Have you tried to do the tutorial first? I have attached one of them for you, but there are many more out there for different versions of BEAST.
Best of luck,
Luther
It's been about 6 years since I used BEAST for my dissertation research, and, frankly, I don't remember what I did to estimate divergence times. The following is the relevant section in my thesis. If it suggests something to you, great! If not, I'm sure others with more immediate expertise will respond:
"BEAST ver. 1.4.7 (Drummond and Rambaut, 2007), employing a Bayesian MCMC coalescent approach, was used to infer the time to the most recent common ancestor (TMRC) for available Lampropeltis triangulum elapsoides specimens. MultiPhyl (Keane, et al., 2007) was used to determine the substitution model and parameters for the 12 L. triangulum sequences in the L595 contig and 97 sequences in the L809 contig. Model settings for estimating divergence times, used the HKY substitution model, empirically determined base frequencies, a Gamma plus Invariant Sites heterogeneity model with 4 Gamma Categories and a strict clock, with Calsbeek’s (2003) upper substitution rate figure (1.79%/my) used to set the molecular clock. The alpha value for the Gamma distribution was set at 0.36. The transition- transversion parameter, kappa, was set at 63.84 and and the proportion of invariant sites was set at 0.7. Data outputs from BEAST were visualized with the associated program Tracer ver. 1.4 (Drummond and Rambaut, 2007). Data inputs for BEAST are formatted through the program BEAUti ver. 1.4.7 (Drummond and Rambaut, 2007). The MCMC chain was set at 10,000,000 steps, allowing for a BurnIn of 10,000 steps."
G'luck!
Mike
It could also be a problem of overparameterization if you don't get the proper values for ESS and other parameters. Since you have a large dataset, I suggest trying with simpler assumptions (e.g., substitution model) and then doing other attempts to assess the variations in the Tracer output. Please also check [email protected] for other suggestions.
Thank you all for your advice and comments. I have gone through the tutorial before starting to work on it. But still had some doubts. Thank you for the older version of the Beast tutorial. It contains some more explanations. I had also posted in the Beast user group. But didn't get a response from there.
@ Zhiqiang Wu, Thank you for kind offer of help. I will send the files across.
- Badges
- Science method
- Similar topics
- Biological Science
- Evolutionary Biology
- Phylogenetics
- Phylogenetic Analysis