Cancel re-ordering of taxa in alignment

[MycoMisfit]

Is there a way to prevent MUSCLE from re-ordering taxa in an alignment?

I see the option is mentioned as the -stable command in the UserGuide for v3.5, but it is doesn't work in v3.8.

I don't see it in the documentation for later versions, so I'm wondering if I have to load v3.5 to prevent MUSCLE from re-ordering the sequences in the alignment.

[rcedgar]

Please see https://drive5.com/muscle/manual/stable.html

[MycoMisfit]

Thank you. I stumbled across this link after I posted my initial question.

Bummer that this issue has infused itself into more recent versions of MUSCLE. I've been using this alignment tool for decades and I don't remember earlier version doing the re-ordering of taxa, nor to I remember encountering alignment issues either.

While I do appreciate putting the like sequences together for better post-hoc manual corrections, I have an old-school method of concatenating datasets to make supermatrices. The re-ordering of the taxa tends to muck this up.

[rcedgar]

Yes, it's unfortunate, and I did put quite a bit of effort into attempting the bug fix, now my focus is much more on MUSCLE v5 and Reseek.

It seems your options are (1) find & fix the bug (post a PR please!) (2) rewind to an earlier version (3) use the python script to post-process.

[MycoMisfit]

going for #3 atm. Seems straighforward enough as long as python is running well on my machine.

[MycoMisfit]

Sadly I've learned a new wrinkle that might just prevent me from using MUSCLE in the future.

To concatenate two datasets, not only do I need the order of taxa to be correct, I will need some taxa with no sequence data as placeholders. MUSCLE appears to be deleting these blank taxa during alignment. I haven't attempted to run the python script, but I'm cynical about it reading two datasets with different numbers of taxa.

[rcedgar]

I guess you mean there are FASTA deflines like >taxon with no sequence. FWIW I would consider that to be invalid FASTA, but this is a de facto standard so different software may handle it differently. I would think about this issue differently. IIUC the goal is to combine several MSAs into a combined alignment using a pre-defined order of sequences. I would recommend writing a python script which combines the MSAs with a supplemental file text containing the taxa in the desired order. Then it doesn't matter if there are missing taxa or the sequences appear in a different order in a given MSA, this is all handled by the merge script. This is a more flexible and robust design which would allow you to choose different MSA software. The merge script is pretty simple, if you don't have strong python skills I would expect a modern chatbot e.g. Claude, ChatGPT 5 or Gemini would very likely write it correctly at first try, and would be able to tweak it to meet any special needs.