Awesome

pirouette_examples

pirouette examples.

Overview

Experiments for the `pirouette` article supplementary material

In the pirouette article, one DD tree is used, from example 30.

For the supplementary materials, certain parameters have been investigated:

30: Use one DD tree
28: use a distribution of DD trees (instead of just one)
31 use multiple artifical trees (instead of just one)
20: the effect of the number of taxa
21: the effect of DNA sequence length
19: the effect of halving a DNA sequence length (that is, from 1k to 500 nucleotides)
34: the effect of double the DNA sequence length (that is, from 1k to 2k nucleotides)
27: the effect of non-clock like models
22: The effect of assuming a Yule tree prior on a Yule tree
26: The effect of assuming a Yule tree prior on a BD tree
23: The effect of differently common diversity-dependent trees
18: The effect of equal or equalized mutation rate in the twin alignment
24: The effect of mutation rate
25: the effect of RNG in the alignment for the error distribution
29: the effect of MCMC chain length on ESS
32 use DD tree with 12 taxa (instead of 6)
33 use 24 taxa (instead of 6)
35 use a mutation rate of 0.25 / crown_age (instead of 1.0 / crown_age)
36 use a mutation rate of 0.50 / crown_age (instead of 1.0 / crown_age)
37 use a mutation rate of 0.75 / crown_age (instead of 1.0 / crown_age)
38 use a mutation rate of 1.25 / crown_age (instead of 1.0 / crown_age)
39 use a mutation rate of 1.50 / crown_age (instead of 1.0 / crown_age)
40 use a mutation rate of 2.00 / crown_age (instead of 1.0 / crown_age)
41 use 32 taxa (instead of 6)
42 use 40 taxa (instead of 6)
43: Use 10x shorter MCMC chain length (from 10M to 1M), sample 1k times
44: Use 100x shorter MCMC chain length (from 10M to 100k), sample 1k times
45: Use 10x longer MCMC chain length (from 10M to 100M), sample 1k times

Other published experiments

Combinations of `pirouette` functionality

These are all runs based on one phylogeny. Each of these examples serves as a high-level test.

Example	Phylogeny	Gen	Cand	Twin	TTM	DSL	STRAF	STWAF	Err
1	Yule	Y	N	N	-	1k	STD	-	nLTT
2	Yule	Y	Y	N	-	1k	STD	-	nLTT
3	Fict	Y	Y	Y	BD	1k	STD	STDEQ	nLTT
4	Fict	Y	N	N	-	1k	STD	-	nLTT
5	Fict	Y	Y	N	-	1k	STD	-	nLTT
6	Yule	Y	Y	Y	BD	1k	STD	STDEQ	nLTT
7	Yule	Y	N	Y	BD	1k	STD	STDEQ	nLTT
8	Yule	N	Y	Y	BD	1k	STD	STDEQ	nLTT
9	Fict	Y	N	Y	BD	1k	STD	STDEQ	nLTT
10	Fict	N	Y	Y	BD	1k	STD	STDEQ	nLTT
11	Yule	N	Y	N	-	1k	STD	-	nLTT
12	Fict	N	Y	N	-	1k	STD	-	nLTT
13	Fict	Y	Y	Y	BD	1k	STD	STDEQ	ADG
14	Fict	Y	Y	Y	BD	1k	STD	STDEQ	LTN
15	Fict	Y	N	Y	CT	1k	STD	STDEQ	nLTT
16	Fict	Y	N	Y	CT	1k	NSL	STDEQ	nLTT
17	Fict	Y	N	Y	CT	1k	NSU	STDEQ	nLTT
18	Fict	Y	Y	Y	BD	1k	STD	STD	nLTT
34	Fict	Y	Y	Y	BD	10k	STD	STDEQ	nLTT
30	DD	Y	Y	Y	BD	10k	STD	STDEQ	nLTT

Legend

Column	Value	Description
Phylogeny	Yule	The true phylogeny is created from a Yule (pure-birth) model
Phylogeny	Fict	The true phylogeny is completely fictious and artificial. If any, it follows both a multiple-birth and a protracted speciation model
Gen	Y/N	A generative model is yes/no hand-picked for this experiment
Cand	Y/N	A set of candidate models is yes/no used in this experiment
Twin	Y/N	The background noise is measured yes/no by using twinning
TTM	BD,CT	Twin tree method, BD=`birth_death`, Y=`yule`, CT=`copy_true`
DSL	[1,->>	DNA Sequence Length in number of nucleotides
STRAF	STD,NSL,NSU	Simulate TRue Alignment function: STD: `standard`, NSL: `node_sub_linked` and NSU: `node_sub_unlinked`
STWAF	STD,STDEQ	Simulate TWin Alignment function, STD: `standard`, STEQ: `standard with equal number of mutations`
Err	nLTT,ADG,LTN	The error statistic used is nLTT, ADG: absolute delta gamma, LTN: Log-transformed nLTT statistic

FAQ

Why are some `TTM` and `STWAF` values empty?

If there is no twinning (Twin = N), one cannot specify a TTM ('Twin Tree Model'), nor a STWAF ('Simulate TWin Alignment Function').

Why are some AppVeyor values empty?

These are empty when there are candidate models (Cand = T), because model comparison is unsupported on Windows.

How are the examples related?

See the table, or use this figure: