Holloway: Fallacy of the Phylogenetic Signal: Nucleotide Level

I found the same thing, see my comment over at Biologos:

2 Likes

I generated new DAG sequences using his link here: Phylogenetic Signal Fallacy: Nucleotide Level - Replit

I haven’t examined his code in the aforementioned link closely (I’m a novice when it comes to python), but it does mention something about leaves, which gave me hope it was outputting sequences of terminal taxa rather than ancestors.

No. He’s just wrong and, apparently, clueless.

2 Likes

Ahh well then it’s a good thing Holloway hasn’t already convinced himself that he’s basically falsified the entire field.

3 Likes

I managed to find a version of his alignment that could be turned into a valid NEXUS file. Did a quick parsimony bootstrap and, as expected, the 50% bootstrap consensus tree had almost no structure. There are a few pairs of taxa with high values (up to 99%), but I’m pretty sure those are just taxa that are almost identical by descent. You really can’t show that non-phylogenetic data are as hierarchical as phylogenetic data if you generate the data using a partially phylogenetic process. The high-bootstrap pairs are taxa 4 and 11, 16 and 18. If anyone knows how these taxa are related, I’d be interested in knowing.

2 Likes

This is what I want to know too. It’d be helpful if there was some easy way to generate a figure like the one Holloway has on his TSZ post.
I suspect they really are generated basically by tree-like descent with a few minor additions or insertions. After all, that does occasionally occur for some taxa when one looks at his figure.

1 Like

The two pairs of taxa in question have long sections of exact match, as one might expect from the algorithm that produced them, and some sections of garbage alignment, as one might expect from Clustal acting on weird sequences. GIGO.

3 Likes

Let’s try and keep this focused on the technical
discussion…

Well, it turns out that all the CI in Eric’s data set was due to poor alignment using Clustal. A correct alignment would have found a CI of “undefined” (0/0) for any tree at all. It was just an example of blindly using programs one doesn’t understand.

3 Likes