We are currently upgrading our qiime1.9/mothur + dada2 pipeline to qiime2.
There is one step in our pipeline where we are aligning our representative asv sequences to Silva database using mothur function align.seqs (see https://mothur.org/wiki/align.seqs/). We are using the default parameters (kmer searching with 8mers and will use the Needleman-Wunsch pairwise alignment method).
From this alignment (over 50 000 bases) we have determined positions and base names for our species of interests.
I would like to find out what would be the alternative of this align.seqs function using qiime2
(I found out this post ; post but it seems not answering my request).
I tried to look for “Needleman-Wunsch pairwise alignment method” or “align.seqs” in the forum but this return no results.
Thanks a lot
QIIME 2 supports a few different methods for aligning sequences. It sounds like you are aligning sequences to build a phylogeny, so you should check out the mafft aligner in q2-alignment, or also see q2-fragment-insertion (which is generally the better way to build a more accurate phylogeny if you have short sequences, see here for more details)
currently, the mafft wrapper only builds a de novo tree, but in the next release of q2-alignment another method will be added to allow aligning query sequences into a reference alignment, which sounds like what you are going for.
Since you have a curated reference alignment, you should check out q2-fragment-insertion (if building a phylogeny is the ultimate goal, not the reference alignment itself per se). Otherwise the mafft-add method in the next release of q2-alignment should do what you need.
No — that is alignment for the purposes of OTU clustering, not building an alignment for a phylogeny.
Let me know if one of those options will work for you!
Hello Nicholas, Thanks ! I ll make tests with mafft aligner in q2-alignment and report here
I have checked q2 alignement [https://docs.qiime2.org/2020.2/plugins/available/alignment/] and to my understanding you cannot align seqences against an aligned database (like silva in our case).
Then I found out a community tutorial: Q2-alignment: reference based alignment using SINA but SINA seems not available in qiime2
Then I found a very interesting topic:Alignment on Reference Database and the user turboqiimer has exactly the same request.
According the above mentioned topic it seems like MAFFT now supports ref based alignments, but those features are not available in the Qiime 2 plugin yet.
I was wondering the following:
- Any idea by when SINA will be available in qiime 2
- Aliternatively does anyone has tested MAFFT (independently of qiime2) to align fasta files on aligned reference database (silva)?
Thanks a lot
As I mentioned above, this functionality is being made available in the next release of q2-alignment, due at the end of this month. Keep your eyes peeled for when that’s out…
Yes, it sounds like that plugin is still under development, I am not sure when it will be made available.
So it sounds like there will be a fix for you in QIIME 2 at the end of this month, but until then you could test out running mafft directly and compare this to your other results.
long story short : I will keep my eye open for the next q2 release and I will in parallel test MAFFT independently of qiime
will keep you posted
short update: I have tested MAFTT independently (using the last version) and so far, I did not obtain the exact same results as the one obtained with align.seqs in mothur.
I am using the aligned silva database over 50 000 positions as it is described here:
https://mothur.org/wiki/silva_reference_files/ (release 119)
Thanks for the update @Francis29029!
As I understand it,
mafft --add and mothur
align.seqs are using different alignment algorithms so it does not surprise me that you do not obtain the exact same result, but functionally speaking I expected you would arrive at similar deliveries (alignment to a reference alignment).
Thanks for opening your other topic about this; since you are attempting to re-build an existing pipeline with QIIME 2 it makes sense that exact is mission critical for your use case, in which case building a new plugin as you plan to do is a great idea.
This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.