My options are "out of sense"

[igel-kun]

I'm trying to simulate a species tree of a bacteria sample, with 100k pop-size, 10k generations and 100 species (taxa), but simphy says that my parameters are "completely out of sense". Why is that? It seems reasonable to me...

Global settings:
--------------------

Trees:
	-Species trees: 1 Birth-death simulations
	        -Speciation rate: Fixed 1.000000e-03,
	        -Extinction rate: Fixed 5.000000e-04,
	        -Outgroup addition: No addition
	        -Stopping rules:
		                -Number of leaves: Fixed 100,
		                -Generations: Fixed 10000,
	-Locus trees: Fixed 3, locus simulations per species tree
	        -Duplication: Genome-wide parameter: Fixed 0.000000e+00,, Rate: Fixed 1.000000e-06,
	        -Loss: Genome-wide parameter: Fixed 0.000000e+00,, Rate: Fixed 5.000000e-07,
	        -Transfer: Genome-wide parameter: Fixed 0.000000e+00,, Rate: Fixed 5.000000e-05,
	        -Gene conversion: Genome-wide parameter: Fixed 0.000000e+00,, Rate: Fixed GG,
	        -Transfer (HGT and GC) sampling strategy: Inversely proportional to the distance
	        -Minimum number of leaves: 1
	        -Minimum number of leaves from different species: 0

	-Gene trees: 1 multilocus coalescent simulations

Parameters:
	-Haploid efective population size: Fixed 100000,
	-Generation time: Fixed 1.000000e+00,
	-Global substitution rate: Fixed 1.000000e+00,
	-Substitution rate heterogeneities
	        -Lineage (species) specific rate heterogeneity gamma shape: No heterogeneity
	        -Gene family (locus tree) specific rate heterogeneity gamma shape: No heterogeneity
	        -Gene tree branch specific rate heterogeneity gamma shape: Genome-wide parameter (hyperhyperparameter) No heterogeneity, locus-tree related parameter No heterogeneity, Gamma parameter(gene-tree related) No heterogeneity
	-Individuals per species: Fixed 100,

Misc parameters:
	-Rooting method epsilon: 0.000001
	-Seed: 3688602402089

I/O options:
	-Output files prefix: bacteria_simphy
	-Verbosity: 1
	-Stats file: OFF
	-Mapping: OFF
	-Database: OFF
	-Parameterization: OFF
	-Command-line arguments: ON
	-Bounded locus subtrees: OFF
	-Output trees with internal node labels: OFF


Simulation:
-----------

Error: Unexpected value has been reached MPFR out of precision!!!! That options must be completely out of sense

Edit: if I lower the value for "individuals per species", the error disappears. So this seems to be a point of contention. But simulating 100 individuals per species already seems very low and I'd consider lower values "completely out of sense" actually....

[adamallo]

Hi Igel,

It has been a while since I have worked on this project, but I will take a look at this and other issues you brought up recently as soon as I can.

From what I remember, MPFR runs out of precision only when a bounded coalescent process is virtually impossible to sample. That usually means that the units are wrong, and you have branches that are way shorter (in coalescent units) than they should be for the number of uncoalesced lineages that reach a locus tree branch that is bounded by a locus-tree event.

It may be helpful to check that all inputs are given in the expected units (especially for time, generations, and coalescent units). Refer to the manual if necessary. It makes sense that when you reduce the number of individuals per species, you ameliorate the issue since you will have fewer lineages to coalesce before reaching the bound. Simulating 100 individuals per species should not be a problem at all for SimPhy.

I will look into this in more detail soon.

DM