qiime phylogeny

Excuse me, I would like to ask what is the difference between the iqtree,fatstree and ramxl when using "phylogeny" to build a phylogenetic tree?

Hi @No.77,

That is a very tricky question to answer as each tool runs slightly different algorithms to generate a tree. You can read more about these tools in our Phylogeny documentation. There is also an associated Phylogenetic Reconstruction video which walks through some of this. In a nutshell, each tool uses different types of tree searching algorithms. Some can be better than others given the conditions

My personal take on this... you need to decide between speed vs a thorough tree search / "robust" phylogeny. FastTree2 is made for very large data sets (e.g. ~ million sequences), and can run very fast on small to moderate data sets at the potential cost of poor tree generation. While RAxML and IQ-TREE 2, are for smaller data sets and are often considered more thorough (again, your mileage may vary). I highly suggest reading this paper by Zhou et al. 2017. There are many papers like this on the subject.

I personally prefer to use IQ-TREE 2 as it will run a model testing algorithm across a few hundred models (this is the default setting in the QIIME 2 phylogeny plugin). After the best fitting model, for your data is selected, IQ-TREE 2 will construct your phylogeny. It can take a long time at this step. But I prefer to generate more accurate phylogenies than making them quickly. Make sure you perform good quality-control and taxonomic filtering prior to tree reconstruction.

Keep in mind, aside from these de novo approaches, there is also a reference-based approach called fragment insertion, which is also included with :qiime2:.

I highly recommend visualizing your tree using empress. I think a tutorial video for this will appear on our YouTube channel soon.

Sorry I do not have an easy answer for you.

1 Like

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.