New type of phylogenetic analysis

[RonO]

https://www.science.org/doi/10.1126/sciadv.adp0114

This paper is trying to identify the genomic changes that occurred
within 5 million years of the Cretaceous extinction event that ended the
lineages of nonavian dinos. They are trying to figure out what changes
occurred during this period of rapid diversification of birds.

They claim to be able to detect changes in base composition of
mitochondrial and genomic sequence. From their own figure 1 I do not
know if they have enough nodes to be able to figure out what changes
occurred during this 5 million year period. You need a lot of branch
points to be able to predict what the ancestral sequence was and you
need them well spaced down through the lineages back to the convergence
that you are trying to figure out what the sequence was. They claim
that they have figured out how to relax the assumption of no change in
DNA base composition as the lineage evolves, and can estimate when base
composition of the genomic sequences was changing.

All they have are the base compositions of the extant genomic sequences,
and they are trying to figure out what the base compositions were in the
past. A neat trick if they can do it since predicting the past sequence
using extant sequences uses the assumption that the base composition was
the same in the past.

Ron Okimoto

[John Harshman]

Post by RonO
https://www.science.org/doi/10.1126/sciadv.adp0114
This paper is trying to identify the genomic changes that occurred
within 5 million years of the Cretaceous extinction event that ended the
lineages of nonavian dinos.  They are trying to figure out what changes
occurred during this period of rapid diversification of birds.
They claim to be able to detect changes in base composition of
mitochondrial and genomic sequence.  From their own figure 1 I do not
know if they have enough nodes to be able to figure out what changes
occurred during this 5 million year period.  You need a lot of branch
points to be able to predict what the ancestral sequence was and you
need them well spaced down through the lineages back to the convergence
that you are trying to figure out what the sequence was.  They claim
that they have figured out how to relax the assumption of no change in
DNA base composition as the lineage evolves, and can estimate when base
composition of the genomic sequences was changing.
All they have are the base compositions of the extant genomic sequences,
and they are trying to figure out what the base compositions were in the
past.  A neat trick if they can do it since predicting the past sequence
using extant sequences uses the assumption that the base composition was
the same in the past.

Not necessarily. There are in fact likelihood models that allow base
composition to vary over time, and even to do so convergently in
different lineages. They can actually help to estimate phylogeny better
when base composition has changed, as when a gene moves from an AT-rich
genomic region to a GC-rich one.