Why to delete gaps in alignments to phylogeny?

Question

[l.souza]

Hello, everybody!

I've been reading the chapter 3 of The Phylogenetic Handbook: A Practical Approach to Phylogenetic Analysis and Hypothesis Testing by Des Higgins and Philippe Lemey, that talks about multiple sequence alignment to phylogenetic approaches.

And the authors suggest deleting gap regions in a post-alignment refinement. But they don't explain why to do this. Could someone clarify me the reason?

Answer 1

[Jean-Karim Heriche]

It is very common to manually refine a multiple sequence alignment after it was automatically generated. The goal is to remove anything that looks dodgy. This can be columns with gaps but not always. One of the reasons gaps are removed is that it may be difficult to compare residues to nothing but this is often context specific. This has been discussed here before. There were also papers on the subject e.g. this one.

[l.souza]

Thank you, Jean-Karim! Would you say that trimAl is trustworthy?

[Jean-Karim Heriche]

No idea, I've never used it. One thing you could do is test how stable your phylogeny is to various refinements. If you get very different results, you should really proceed carefully.