Death rate higher than birth rate

[Arthur Zwaenepoel]

Hi,

I would love to use SimPhy to simulate locus/gene trees for a fixed species
tree under a large variety of parameter values. I would like to use maximum
likelihood estimates of the loss and duplication rates which I obtain using
gene count data (as in e.g. Hahn et al. (2005), Rasmussen & Kellis (2011),
Librado et al. (2011) or Rabier et al. (2014)). These estimates result almost
invariably in a higher loss rate than birth rate. However SimPhy seems to
complain when trying to use a loss rate which exceeds the birth rate:

Settings error: : improper sampled values

Is it impossible to have a larger loss rate than birth rate in the current SimPhy
implementation? I can imagine a workaround, but it is not very convenient.
Excuse me if this question is redundant, I was unable to find a clear answer
in the manual.

Best, Arthur

[Arthur Zwaenepoel]

Hi,

I would still love to use SImPhy for my research purposes. I have looked into
DLCOAL_sim in the meantime, which also kind of serves my needs, however
lacks the advanced branch length heterogeneity features. Still no advice on the
use case with higher death then birth rate when using a fixed species tree?

Best,
Arthur

[Diego M.]

Dear Arthur,

I have just added a flexibility option to solve this problem and committed it to a new branch in SimPhy's GitHub repository (flexiblesim, https://github.com/adamallo/SimPhy/tree/flexiblesim). You just need to add "-f 1" to your input parameters (either in the command line or input file).

I have not merged with master yet since I need to document this new option on the wiki. Let me know if you need help compiling the sources.

SimPhy's algorithm for locus tree simulation allows loss rates that are smaller than birth rates. However, this is not true for the species tree, and there was an inherited setting check-up that is not strictly necessary. However, loss rates that are much smaller than their birth counterpart will generate simulation conditions in which a lot of simulations are not valid (extinct trees). SimPhy tries to avoid this problem repeating the simulation of locus trees until it generates the number of desired locus trees. This should not lead to an infinite loop since SimPhy has a maximum number of iterations in place but may lead to much longer running times. If this happened, you could see it using verbosity >=3. Example of the warnings:

Locus tree with %u leaves, less than the minimum %u , restart of the simulation of this tree. Try %d of %d 

Locus tree with %u leaves from different species, less than the minimum %u , restart of the simulation of this tree. Try %d of %d

Locus tree with %u associated gene tree leaves, less than 2, restart of the simulation of this tree. Try %d of %d

I hope this helps,

            Diego M.

[Arthur Zwaenepoel]

Hi Diego,

Thanks a lot! I will test it out somewhere in the near future!

Best, Arthur

[Arthur Zwaenepoel]

Hi Diego,

I succeeded in building the flexiblesim branch, and it works like a charm.
Thanks a bunch!

Best, Arthur