I have been examining some UCE datasets deposited in Treebase, and the alignments, as far as I can see, are very poor, with floating small fragments that are obviously "aligned" at random, and clear mismatches in gaps to which more parsimonious alignments are readily apparent. This seems unacceptable to me, but these datasets are published, so I guess editors and reviewers think this is not a problem. Is it? Should it be?
I have looked at a few UCE datasets in TreeBase, and I agree with you, they look like a lot of low quality data. I think the UCE approach sounds good in theory, and the end result trees in the publications look quite reasonable. So I assume there is just a massive amount of signal along with the noise in the data sets, so the sum total of the data works, even when each of the alignments looks rather poor.
If that is the case, then what is the criterion for determining that it "works"? That it gives you back a tree that agrees with one you already had from some other data set?
- Badges
- Science topic
- Similar topics
- Filing