Using reverse reads only?

[V.Hale]

Hello!
I recently received sequencing data in which the forward and reverse reads were switched. The R1.fastq.gz files are actually reverse reads while the R2.fastq.gz files are forward reads. This was determined by looking at the rep.seqs and seeing that all R1 files start with the reverse primer while all R2 files start with the forward primer. The R1 files (which are actually reverse reads) have much higher quality than the R2 reads. I would like to try using single end reads (not paired) due to the poor quality of the R2 files, but I have a few questions about this:

1. Would renaming the R1 files "R2" and the R2 files "R1" change anything about how the files are imported into QIIME2?
2. If I remove the R2 files and only use the R1 files (which are actually reverse reads) for single end analysis, and import them as "R1" files, will I get accurate taxonomic assignment? (They are being recognized as "forward reads" in the demux file, but they are actually reverse reads.)
3. If I rename the R1 files "R2" and import them as "reverse reads" (if this is possible), could I use reverse reads only in an analysis?

I am in the process of trying all of the options above, but am not necessarily sure what's happening under the covers and if one of these options is more sound than the others.
I appreciate any guidance that can be provided!
-Vanessa Hale

[Nicholas_Bokulich]

Hi @V.Hale!

@V.Hale

Would renaming the R1 files “R2” and the R2 files “R1” change anything about how the files are imported into QIIME2?

No. Note that you could also use the single-end manifest format, then instead of changing the name of the files you can just list the filepaths that you want to use in the manifest file.

@V.Hale

If I remove the R2 files and only use the R1 files (which are actually reverse reads) for single end analysis, and import them as “R1” files, will I get accurate taxonomic assignment? (They are being recognized as “forward reads” in the demux file, but they are actually reverse reads.)

As long as the reads are all in the same orientation in the file, this should not impact the classification results. The classify-sklearn method can autodetect orientation, and the other classifiers in q2-feature-classifier can be told with orientation to test vs. the reference sequences.

@V.Hale

If I rename the R1 files “R2” and import them as “reverse reads” (if this is possible), could I use reverse reads only in an analysis?

No... because right now the formats used are single-end (forward) or paired-end. As noted above, though, the R1/R2 naming convention in CASAVA 1.8 is not totally binding and is ignored if you use the manifest format... in which case you specify manually what each file is.

Good luck!

[V.Hale]

Thanks so much for this information! Will try the manifest format.
Cheers!