Nested Clades, The Consistency Index, and Affirming the Consequent

[EricMH (Eric Holloway)]

@evograd

That’s the biggest one to solve up front. After you’ve done that, we can look and see what the other problems might be.

Alright, I’ll make that adjustment next.

@evograd

Do you understand why you can’t compare results from Klassen’s binary datasets with these nucleotide datasets? Playing around with different datasets it appears that simply going from a binary dataset to a quaternary dataset, keeping everything else the same, produces ~2x higher CIs.

Maybe I don’t understand what you mean by a binary morphological character. I think you mean something like ‘has wings’, which for some creatures is true and others is false. We have an array of such properties, and each creature gets a string of trues and falses, or 1s and 0s, representing whether they have each particular property. If this is what you mean, then as I said, this is the first analysis I did, and the CI values were even higher than what I get with this DNA version.

You can see that work here.

Fallacy of the Phylogenetic Signal? Part 2

Update, I had an error in calculating the normalization factor for the CI metric, so here are the new results: Average adjusted CI over 100 runs: evotree: 0.8676219452202528 dag: 0.7991580076656275 fake: 0.39394993245276183 And when comparing to the null hypothesis: Tree hypothesis falsified 0 / 100 times. DAG hypothesis falsified 17 / 100 times. Fake hypothesis falsified 52 / 100 times. Makes more sense! Still shows CI is lacking when it comes to inferring evolution. Fake leaves stil…

@evograd

Can you post the original data you used there?

I used this primate dataset: http://paup.phylosolutions.com/wp-content/uploads/2018/02/primate-mtDNA.zip

Which I got from here: Quick Start | PAUP* (* Phylogenetic Analysis Using PAUP)

@evograd

I certainly don’t get well-supported branches when I run bootstraps of DAG-derived trees.

Can you post your settings so I can reproduce what you’ve done? For some reason I can consistently get well supported trees with my datasets. My settings are probably wrong.

Also, the question is not ‘are there DAG datasets that produce bad trees?’ Rather, the question is ‘are there DAG datasets that produce well supported trees?’

Some of the DAG datasets are indeed pretty poor, but others seem to produce well supported trees.

@evograd

It’s also worth noting that I consistently failed to get a high rescaled consistency index (RCI) with the DAG-derived trees.

This may indeed be the fatal flaw with my approach. I don’t know what counts as a ‘high’ RCI value. Do you know the range of ‘high’ RCI? The values I see from my datasets are pretty small, less than 0.1.