search

tschuelia / PyPythia

Lightweight python library for predicting the difficulty of alignments in phylogenetics
GNU General Public License v3.0
16 stars 0 forks source link

The predicted difficulty for MSA is: 0.6 #14

Open tschuelia opened 9 months ago

tschuelia commented 9 months ago

Referencing from issue #13 :

Question by @liamxg

Dear @tschuelia, if the predicted difficulty for MSA is: 0.64, it is very hard or hard?

tschuelia commented 9 months ago

Hi @liamxg, I opened a new issue for your new question. Please don't change the title and text of an issue to make sure other uses who might have a similar issue can find the solution as well 🙂

Now to answer this question: we would classify an MSA with difficulty 0.64 as intermediate. This MSA is not trivial to analyze, and you should be careful to report a single tree as the optimal tree topology. Instead, you should run a thorough tree inference with multiple starting trees (i.e. different initializations of the tree search), run additional statistical tests (e.g. as implemented in IQ-TREE) and check the bootstrap support of your tree and maybe summarize the results in a consensus tree. As for the thorough tree inference: you can use our latest version of RAxML-NG (https://github.com/togkousa/raxml-ng/tree/adaptive developed by @togkousa), that implements a difficulty-aware search heuristic. This heuristic determines the best tree inference strategy (and number of starting trees) based on the difficulty of the MSA. You can read more about this in the respective publication in MBE: https://doi.org/10.1093/molbev/msad227

liamxg commented 9 months ago

Dear @tschuelia, That's helpful.