So I am working with Cytochrome oxidase I and using it to identify species within a genus. I built a tree with Mrbayes and retrieved a good tree with highly supported nodes.
However, I was advised against using identical sequences belonging to different specimens so I removed them and built the tree again only to retrieve nodes with very low values. The species form the same clusters but are not well supported (some nodes are as low as .32).
It's important to mention that I am not trying to build a phylogenetic tree, I only wish to see how the specimens cluster within a tree, so we can "count" the number of species present and later on review their morphology.
Any help is appreciated!
Hi Luz!
Based on your goals, I don't think you should need to get rid of them. Are they 100% identical (have you tried looking for nucleotide differences between them)?
I am not sure phylogeny of any one small gene like Cytochrome Oxidase I can be used to determine species within a genus, but it depends somewhat on the type of organism you are studying (plant vs fungi vs animal).
Hi Luz:
One reason to not include identical sequences is to do a faster analysis. In your case, I agree with Jose Alberto Lopez-Aleman you should keep those identical sequences. Since your interest is having a tree to help in the identification of your specimens, I would also recommend you to use a Maximum Likelihood approach such as RAxML or IQtree; they are faster than MrBayes and you do not need to worry about priors.
I hope it helps!
You could try using programs more focused on delimiting species, for example STACEY, BPP, bPTP or TR2. They all use Bayesian analysis, but accept reference trees as input to decrease the time it takes to get results. Maybe if it's just COI, don't delay that long.
Multiple, identical sequences produce polytomies in cladistic analyses, or spuriously resolved nodes in other methods. While it is true that one-gene phylogenetic hypotheses are rather minimal these days, that's what the Barcode of Life folks do every day. Remember, Bayesian inference is subjective - it is about how much you believe your results, not about how much I should!
- Badges
- Science topic
- Similar topics
- Journal Impact Factor