Same classifier, same otu, different sequencing run = different classification

[Nicholas_Bokulich]

@shira

If I understand correctly, what you suggest I do is process the raw data as you would normal paired-end reads, and only change the way the taxonomy is determined

Correct

@shira

Is it OK to assume that dada2 will deal well with the situation of mixed reads?

Yes

However, this will inflate the number of features, altering (especially) alpha and (probably) beta diversity results based on ASVs. If that's not a problem for your experimental goals, then there's nothing to worry about.

If that is a problem, reorienting your reads prior to importing to QIIME2 may be a better choice, e.g., using the script in the post I linked to above. Or wait until we provide support for this in QIIME 2 (we will post here when that feature is added in a future release).

This thread may also contain some useful advice for now.

@shira

Would you try and avoid a company that provides the raw data in mixed orientation format, if so, what are the reasons?

Not necessarily. There is nothing "wrong" with what they are doing, it's just unusual and hence we don't have the functionality required to process those reads in a streamlined way in QIIME 2 . We will in the future because a few users have asked

@shira

Following your suggestions I tried using vsearch. I could not find a tutorial to follow, so I came up with the following code. It’s still running, I hope it works

Yes, that should do it . If it's still running, it is still working! It will cause an error if you run out of memory or some other issue, so don't worry for now.

@shira

have a backslash \ at the end of each line, but its not showing here for some reason

put code between lines containing three backtick (```) characters to make it appear like code on the page, and prevent reformatting.

I hope that helps!