search

xavierdidelot / BactDating

Bayesian inference of ancestral dates on bacterial phylogenetic trees
https://xavierdidelot.github.io/BactDating
MIT License
82 stars 15 forks source link

calculation of mutation rate and DIC value #49

Closed LeonardosMageiros closed 3 years ago

LeonardosMageiros commented 3 years ago

Hi,

Running bactdating on ~1500 S aureus strains and printing the result gives me the following: Rooted; includes branch lengths. Probability of root branch=0.82 likelihood=-7.83e+03 [-7.90e+03;-7.75e+03] prior=-1.28e+04 [-1.30e+04;-1.27e+04] mu=1.91e+00 [1.67e+00;2.15e+00] sigma=1.59e+01 [1.41e+01;1.76e+01] alpha=5.97e+03 [5.14e+03;6.94e+03] Root date=-503.78 [-908.13;-148.30] Root date for most likely root=-532.65 [-911.61;-203.64]

Can I ask what are the units of mu how I can calculate the mutation rate (per year) from the mu value?

Can I also ask how can I interpret sigma, alpha and DIC values? Should I be alarmed that my DIC value is 20878.84? Is that too high?

Thank you in advance for your help, Leonardos

xavierdidelot commented 3 years ago

Hi Leonardos,

mu is in substitutions per year.

sigma is harder to interpret and depends on the model used, but the larger the sigma the more relaxed the clock. I would not usual report this value other than to say that the clock used is a relaxed clock. Here it seems you are using the new ARC model which it would be good to mention. The paper on the ARC model which explains exactly what sigma is: Didelot, X., Siveroni, I., and Volz, E. M. 2021. Additive uncorrelated relaxed clock models for the dating of genomic epidemiology phylogenies. Molecular Biology and Evolution 38:307–317.

alpha is the effective population size, which is not especially interesting to report since it is reflected directly in the root date.

A high DIC value does not mean much by itself, the actual value is only meaningful if compared with other DIC values for the same dataset, for example using a strict clock model.

Best wishes, Xavier

LeonardosMageiros commented 3 years ago

Xavier thank you very much once again.

Just to be 100% sure mu is measured in substitutions per megabase per year, correct?

Thx

xavierdidelot commented 3 years ago

No, mu is in substitutions per year for the whole genome. If you want it per year per bp you should divide by your genome length.