Hello, I have a question about 'SampleData[SequencesWithQuality]' data.
I tried to merge two 'SampleData[SequencesWithQuality]' data (one from single-end sequences and the other from paired-end sequences after read joining) before using 'demux summarize' to summarize counts per sample but failed to find a way.
Is there any way to merge two 'SampleData[SequencesWithQuality]' data?
Thanks for reading!
Hey @Jay ,
No, basically because there is usually not much cause to (more below)*.
You definitely cannot do this, since they are different types! The latter would be
SampleData[JoinedSequencesWithQuality] (if joined) or
SampleData[PairedEndSequencesWithQuality] (if unjoined), not
*Why this is usually not needed:
you could merge, but in most cases it would not be beneficial, or might even be a bad idea. If denoising reads with dada2, the datasets should not be merged upstream, because dada2 builds its error model on individual sequencing runs, so these should be kept separate. If using deblur for denoising, or if using some types of OTU clustering (closed or open-ref), you can still denoise/cluster runs individually, and it probably won't impact results (with open-ref you need to pass in the "new reference sequences"/reference + OTU centroids from the previous run to make sure you use the same OTU centroids and IDs, but otherwise the same advice applies).
In all the former cases, just denoise/cluster and then merge the outputs downstream before proceeding with analysis.
If using de novo OTU clustering, this is one case when you need to merge prior to clustering... but in that case I would recommend still using some type of denoising (which will output
FeatureData[Sequence] artifacts, which can be merged with
qiime feature-table merge prior to de novo OTU clustering) since de novo clustering does poorly with erroneous sequences.
So unless if you want to party like it's 2011 (de novo clustering without denoising), there is usually no need to merge in advance.
But if you have a special use case this is not covered here, just let us know! (and if you really really want to merge these types, you could always contribute a
merge action to q2-demux!)
Thanks for your kind and detailed reply!
It really helped me a lot.
This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.