I'm analyzing 445 protein sequences of a single protein for phylogenetic analysis. Multiple sequence alignment of these sequences is showing gaps of different sizes. Is it ok to keep these gaps for tree making or I should remove them before submitting to tree building.
If you are removing gaps, you are messing with MSA. So not, do not do anything manually. If you want to remove something, read about TrimAl or Gblocks program.
I would perform the phylogenetic analysis on the full MSA without gap-stripping. I would do same for the gap-stripped dataset. In this way, you can easily check if gap-stripping will confound the tree topology.
Since you are handling a large dataset, consider excluding the gappy sequences from the analysis. In rare situations where I find lots of gaps caused by a single or very few sequences, I would exclude this sequence(s) especially if they are reference or database sequences, or delete the entire column. Either way, always compare the tree topology to the original full dataset.
Hello,
Gaps in multiple sequence alignments are placed so that residues present in the same site/column are (supposely) homologous, this is a prerequisite of phylogeny inference. Hence, as pointed out by Abhijeet Singh you have to be cautious when modifying the alignment.
Several tools allow to automatically filter out the unreliable part of sequence alignments (by removing site/sequence or masking residues) in a reproductible way. The relevance of such filtering methods, before phylogeny inference, has been questionned in a paper of 2015 (Article Current Methods for Automated Filtering of Multiple Sequence...
). However this paper focused on filtering tools that can only remove a whole site or sequence from the alignment (such as trimAl or Gblock) whereas some alternative filtering methods (such as HMMcleanerArticle Evaluating the usefulness of alignment filtering methods to ...
or guidanceArticle GUIDANCE2: Accurate detection of unreliable alignment region...
) are also able to mask indivual residues. In my experience HMMcleaner do a really goog job, I don't have much experience with guidance.- Badges
- Science topic
- Similar topics
- Filing