Error Muscle alignment (msa package, through Rscript)

Question

[alnam]

Hi,

I have encountered a problem, while running Muscle through the MSA package in R (via Rscript). I get this:

*** ERROR ***  MSA::SetIdCount: cannot increase count
Fatal error, exception caught.
Error in msaFun(inputSeqs = inputSeqs, cluster = cluster, gapOpening = gapOpening,  : 
  MUSCLE finished by an unknown reason

And I couldn't find what it meant. Is it that I have too many sequences..? If anyone has an idea of how to solve this, it would be great.

Thanks!

[Joe]

What is your input data like? You haven't told us how many sequences you have so we can't tell you if its too many.

Have you tried running it on a subset of your data?

It would also be worth trying to run the data through MUSCLE directly, without the R wrapper.

[alnam]

Thanks for your answer! I am running it on 10,000-20,000 sequences of around 1,500bp, 96 times. When running individually a case that caused a problem in the loop, there was no issue. I think the issue was what is described in the section "Known issues" of the msa package (here: https://bioconductor.org/packages/release/bioc/vignettes/msa/inst/doc/msa.pdf ), which is that there can be memory leaks with Muscle? Anyway, I used ClustalOmega instead and it worked very well.

Thanks again!

[Fatima]

Are your sequence headers unique? Are you using a profile? Could you provide the command that you're using?

This is the part of the code that gives you the error:

void MSA::SetIdCount(unsigned uIdCount)
        {
        if (m_uIdCount > 0)
            {
            if (uIdCount > m_uIdCount)
                Quit("MSA::SetIdCount: cannot increase count");
            return;
            }
        m_uIdCount = uIdCount;
        }

If you search MSA::SetIdCount in the link below, you can find the codes where this function is used, and it might give you some ideas of the reason behind the error:

https://git.wur.nl/haars001/reas/-/tree/master/muscle3.6_src

[alnam]

Hi, thanks for your reply!

Yes I had found this, but was not sure of what it meant (what is m_uIdCount ?). Yes my headers are unique (but I wasn't sure there was a problem if not, so thanks for this) and I don't know if I am using a profile. As I said, I have found a solution, which is using clustalomega instead of muscle. I agree it does not solve the issue but at least it helps.

Thanks again!

[Mac]

hi, did you manage to resolve this? I am having the exact error. I created an alignment with 2067 sequences using the "dna = msa(dna, method = "Muscle", order="aligned")", and the input file was a "DNAStringSet". I then moved to do the second alignment which had 174 sequences and it also ran okay. But when I moved to the third alignment with 2644 sequences, I received the error "* ERROR * MSA::SetIdCount: cannot increase count". So on smaller alignments it seems to be running okay and only get the error when i move to bigger datasets

[Mac]

I have changed the method and used "dna = msa(dna, method = "ClustalOmega", order="aligned")" instead and it worked okay. I think there is a bug in the "muscle" method

[alnam]

Hi (quite late), sorry I had completely forgotten this. I did the same, there seems to be some issue with the muscle method (I haven't checked if it is still the case, though). Thanks for sharing your solution here!

[gildas.lepennetier]

I have this problem with 7 sequences.

SEQS
"A DNAStringSet instance of length 7"
names(SEQS):
"1" "2" "3" "4" "5" "6" "7"
any(duplicated(names(SEQS)))
"FALSE"

all sequences have a width of 361 nt.

Those sequences are also not the same:

require(stringdist)
d=stringdistmatrix(a=as.vector(SEQS), b=as.vector(SEQS), method="hamming",useNames ="names")

d[1,]
"1  2  3   4  5  6  7"
"1  0 16  6 22 13 17 12"

As previously noticed, it is working using "method="ClustalOmega" (using Gonnet ??)

Answer 1

[ulrich.bodenhofer]

Just because of a recent issue raised on the package's GitHub page (https://github.com/UBod/msa/issues/30), I became aware of this problem and tried to solve it. I am sorry for not having seen this one here for more than four years (!!), but I hope the solution still helps in some way.

The point is that there was a major bug in the Muscle interface that - to my deepest regret - had not been discovered by us so far: If you run Muscle more than once, you can only run it with no more sequences than in the first call. The solution was not that trivial, but I tried my best. A new version (1.37.3) has been pushed to the GitHub repo (https://github.com/UBod/msa) and the BioC devel branch now. I hope this closes this issue. If you encounter any further problems, please let me know by reopening this issue on GitHub (https://github.com/UBod/msa/issues/30) or opening a new one.