Hello,
suppose I have constructed a phylogenetic tree with all OTU sequences and reference sequences. I now want to have a tree containing only the OTU sequences. Is it a good practical to remove the reference sequences from the original tree? Will such handling interfere the topology/clustering of the OTU sequences.
I read somewhere in web stating "Deleting nodes from phylogenies is risky business! Deleting nodes is unlikely to give you the same phylogeny as re-running the tree with the corresponding sequences removed." how will node deleting interfere with the clustering alone? Are there relevant literatures on this matter?
Thank you in advance and best regards
There is not enough information to give any exact suggestions. But in general, any removing or adding taxa into a phylogenetic tree can affect to its topology. These changes are in relation to math model used for a phylogenetic tree construction, general number of OTUs and general variability changes in sampling caused by such deletion.
Hello,
taxon sampling is known to affect the accuracy of inferred phylogenies. The main reason is that additional taxons may break some long branches of the tree and reduce problems caused by homoplasy
http://materiais.dbio.uevora.pt/Evo/A_review_of_long-branch_attraction.pdf
http://www.plantsystematics.com/qikan/manage/wenzhang/jse08016.pdf
http://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.1000602
I hope this help,
sincerely
Hi Vincent,
Nice papers to get some understanding. Why I have the impression that more sequences / taxa yield better phylogenetic inference. The problem I am currently dealing with is OTUs of some bacterial functional gene homologues. The dataset has been processed by mothur and contains OTUs with at least 5% sequence differences between each other, and reference sequences from NCBI.
provided my impression is right, it should be ok or even better to construct phylogenetic trees will All available sequences and then prune the tree down to one tree containing only the sequences/taxa that I am interested. Am I right or I get it wrong?
Best wishes
Hello Yiyu,
I deem that your suggestion to use all available sequences for the construction of the phylogenetic tree and then prune the outgroup (reference) sequences ia a good idea. Outgroups are very important in the construction of a tree; if you do not include them in the phylogenetic analysis, the overall topology of the tree will change.
If you are using Dendroscope for tree visualization, then you can simply collaple the "outgroup" node.
However, you can try re-constructing the tree using different sequences and see what you get.
Regards
Hi Yiyu,
An important point you do not mention is why you want to do that? I agree with all of the above that the broader and more diverse the taxonomic sampling, the better the chance that your inference will correctly reconstruct the phylogenetic history of your gene (notice that this does not necessarily corresponds to the history of the taxa carrying the gene). If you want to prune out some of the leaves/tips of your final tree for visualization purposes, that is perfectly fine, and several tree-processing packages will allow you to do it without changing the topological relationship among remaining leaves/tip. Just make sure you say explicitly that you did so in the methods of the eventual paper and in the caption of the figures containing the pruned tree.
On the other hand, and as Athanasia points out, removing sequences and reconstructing the tree should also give you an idea of how robust is the topology you are getting to changes in the input. So you should experiment with it, so you can see for yourself how removing certain sequences can greatly change the resulting tree, while others may have little or no influence at all. You will gain a good amount of insight if you take the time to do it, and also gain more overall confidence on your final results.
- Badges
- Science method
- Similar topics
- Biological Science
- Evolutionary Biology
- Phylogenetics
- Phylogenetic Analysis