No. But many phylogenetic programs treat the gaps as missing information, some treat a gap character as a 25% match to any of the four bases, some treat a gap as a mismatch.

There is not good way to assign a value to gap characters because different types of insertion and deletions are common. An addition of another repeat in a repetitive element for example is VERY much more common than addition of random DNA in a random site. Addition or deletion of 3 or 6 or 9 bases in a row is probably a single event and not 3 or 6 or 9 independent mutations. Deletion of a larger chink of DNA such as a transposon is also a single and common event.

If you really want to include in/dels in a calculation it is possible to recode the gap characters as other characters and then teach your phylogenetic program how to treat them. For example, insertions or deletions of one or more copies of a repetitive element could get an "x" character in place of the "-" gap character, wile single base deletions are replaced with "z". But deciding how much weight to give to each type of insertion or deletion may be difficult.

In general, if you are not sure how the program you are using treats the gap characters, you can gap-strip the data first (by column in the alignment, using a gapstrip program, not be deleting "-" from the file which misaligns the data matrix). I recommend building tree with and without gap-stripping to compare the results.

https://www.hiv.lanl.gov/content/sequence/GAPSTREEZE/gap.html

Thank you so much for the response Brian,

Fortunately I am dealing with relatively short haplotypes with good sized gaps (I've included an example of one case below), so I feel like they are valuable to code as meaningful evolutionary events that are differentiating populations of the same species.

AC - - - - - - - - - - - - - - - - - - - GCTGC

AC T T C A T A G A T A A C T A G T T A G GCTGC

The models I've tested thus far all remove the gaps completely so needless to say I am exited to try GapStreeze. I will provide feedback once I have had the chance to test it out.

Thanks again.

I see now that GapStreeze just removes the gaps yes? That doesn't actually help me much as I can easily remove the gaps from the alignments in CodonCode Aligner and then export them as a gap-less alignment for analysis. I'm sure removing gaps in R would be pretty simple too, but why remove them if the distance models ignore them anyway?

So I rephrase my question. Can anyone point me in the direction of a tree building method that takes SNPs and in/dels into consideration when developing a distance matrix (I really just want the distance matrix actually)?

Thank you in advance.

GapStreeze can be set to remove columns with any user-defined percentage of gap (or non ACTG characters, you can use any list of characters you like there) characters in it. This is good for removing columns where most species have a gap or missing information but not loosing columns where only a few are represented by gap. You gave an example of just 2 sequences, so it is only a wild guess as to whether one species has a deletion where most have sequence that is homologous, or whether one has an insert and most species do not have an insert there. In the case of insertions, you can have either insertions that are homologous (the common ancestor of some species experienced a one-time insertion) or non homologous (insertions tend to happen at that site for some reason, but the inserted DNA is not fro the same source each time).

If you think the deletion is quite "important" relative to single DNA base changes, you can for example replace the "-" characters with "T" characters and now all the species with the deletion are identical there but different than the ones that have not had the deletion. If you think the deletion is a rather minor event, you could replace the dashes with the consensus of the sequence that were deleted and maybe change one character so now the species with the deletion are similar but not identical to the species that did not have the deletion, but they are all identical to each other because we claim that the deletion was a one-time minor event.

Build the tree with and without these types of fudging with the data, to see how much difference it makes. You can also use the data as is, and try programs that treat gap characters in different ways, but as far as I know none of the programs has built-in intelligence to say "this is a one-time insertion of ten bases so we treat it as such; while another insertion seems to be a site where transposons often come and go leaving tidbits behind, so we treat that differently".

Brian Thomas Foley I replaced the in/dels with a/t's and it does make a big difference in the trees produced. Do you know of any publications where this or a similar approach to treating in/dels was practiced?

I am not aware of any publications where it has been done. But if it is a good idea, someone has to be the first one to do it. Any reviewer could argue the merits of this (or why it should not be done) in theory, forever. So if you think it helps your analysis to include the in/del data then I would recommend taking the same region of the genome from a set of organisms where the "true" phylogeny is well know from fossils, complete genome work, or whatever. Then if using this small region with in/dels treated with your method "improves" the trees toward the "true phylogeny" for that group of organisms, you have a strong argument that the same method is valid on your organisms which I presume have an as yet unknown "true phylogeny".

For one example, I have built phylogenetic trees using complete mitochondrial genomes of vertebrates (mostly tetrapods, with fish as the "outgroup"). Some of those phylogenies looked very bad given that we know for sure that amphibians, reptiles, birds and turtles have been roaming the land for longer than mammals. But I found I could improve on that by improving the multiple sequence alignment, deleting some unalignable regions, using better partitioning, etc... So now when I use complete mitochondrial genomes of insects, I can be more confident that I know how to handle the data.

Using just mitochondrial D-loop, or other small regions of mitochondrial DNA for all vertebrates makes a real mess. But the same regions are quite reasonable for groups with far less diversity such as looking at species and subspecies of one genus of bats, or domestic goats or sheep. So you need to pick a data set that is comparable to whatever you are looking at. If you are looking at fungi which have evolved over hundreds of millions of years, then the "all vertebrates" data set may be what you need, but the answer may be that it takes a lot of good data to build a good tree across that range of distances, and that re-coding the gaps does not help much.