Hi, I was initially get confused on why my denoising result only showed very few sequences that passes filter. Then I checked my demux output and it has very short sequence length. I used v3-v4 and was sequenced on 2x300bp.
Is this normal or my reads are basically not good? If it's not good, is there any way that it's salvageable?
This is the import command that I ran :
qiime tools import --type 'SampleData[PairedEndSequencesWithQuality]' --input-path fastq --input-format CasavaOneEightSingleLanePerSampleDirFmt --output-path 1_demux.qza
I also attach my demux results here
1_demux.qzv (321.5 KB)
Any ideas?
thank you 
Hello @cmetadea,
Yes these look less than ideal. What quality control steps were performed on your reads before importing into qiime?
Hi Colin,
I imported the reads to qiime directly from the sequencing output. Am I supposed to do something first beforehand?
Thanks
Hello @cmetadea,
No not necessarily, but because the reads are such variable lengths I assumed that some sort of processing had been done. I would reach out to whoever performed the sequencing, as they may have performed some primer or quality score trimming that apparently went a little off the rails.
Hi Colin,
We did the sequencing in-house and this was the output sequences directly from the machine (we left it default on the illumina machine).
Does this mean that if I decide to proceed, very few reads will be retained and I only got few left to do my analysis?
Thanks
Hello @cmetadea,
You have two options: try to merge the paired reads and see what you end up with, or proceed with only the forward reads. I would try both and see which gives you better results.
Does this mean that if I decide to proceed, very few reads will be retained and I only got few left to do my analysis?
If you go the paired end route, I think that yes, you will have few reads to analyze compared to what you started with. If you go the single end route, the number of reads won't so much be the problem as will the length of the reads.
Hi Colin,
if I only use forward reads, will it affect the trimming? (e.g. can I just use regular v3v4 trimming as I did for paired end or just v3?)
also, for taxonomy assignment, should I train new ones with only v3 region?
Thanks
Hello @cmetadea,
if I only use forward reads, will it affect the trimming? (e.g. can I just use regular v3v4 trimming as I did for paired end or just v3?)
Are you referring to primer trimming or positional trimming and truncating in dada2?
also, for taxonomy assignment, should I train new ones with only v3 region?
That's a good question, I'm not sure. It seems like this would be a good idea but there might be some gotchas since you amplified the v3+v4 region and not only the v3 region. I would do some research on the forum and open a separate thread if you don't find anything.
Hi Colin, sorry for the late response.
I referred to the primer trimming.
I will indeed try for tax assign using v3v4 and will update you if I find anything.
Thank you