Losing almost all reads after running “qiime vsearch join-pairs“” on paired-end sequences

[zy-yuan (yz-yuan)]

hi,everyone.I have 206 samples in total，and i imported my data as the following.

qiime tools import
--type 'SampleData[PairedEndSequencesWithQuality]'
--input-path jej.csv
--output-path jej-paired-demux.qza
--source-format PairedEndFastqManifestPhred33

It is run successfully,so i want to have a look about my data description. i run the command as following.

qiime demux summarize
--i-data jej-paired-demux.qza
--o-visualization jej-demux.qzv
jej-demux.qzv (291.5 KB)

#I put the detailed info in the printscreen.I have 12787699 frequency sequence ,and per sample about 62076 sequences.
after that,i make a joint of my forward and reverse sequences.

qiime vsearch join-pairs
--i-demultiplexed-seqs jej-paired-demux.qza
--o-joined-sequences jej-demux-joined.qza

#and then,change the result form into .qzv have a look.

qiime demux summarize
--i-data jej-demux-joined.qza
--o-visualization jej-demux-joined.qzv
jej-demux-joined.qzv (293.6 KB)

but the result is amazing,only 4269 sequences reserved totally.I have no idea about this result ，and why almost sequence are removed.
and I have not run the "qiime deblur denoise-16S".
Any help is much appreciated!

[jwdebelius (Justine Debelius)]

Hi @zy-yuan,

Based on your quality plots, it looks like your reads are very low quality, especially as you go past 150 bp in both the forward and reverse reads. Due to the poor quality, Vsearch many not have been able to get a good consensus sequence and merge the reads. I think if you look at the vsearch log, you can see how many reads were dropped.

My suggestion for this sequencing run would be to use the forward reads alone, and trim them between 120 and 150 bp.

Best,
Justine