Optimal proportion of non-chimeric reads in my samples

Hello all,

I am new to qiime. I have 2 NGS runs of 50 and 43 human stool samples respectively. In one of the run (43 samples), after trimming primers, adaptors etc. I ran DADA2. However, I can retain ~60-80% of my reads per sample (which are flagged as non-chimeric). I have read different posts in this forum and realise that 60-80% is not that low, but should I play a bit with the --p-min-fold-parent-over-abundance parameters or continue with the retained reads? I am getting ~2300 sequence features from DADA2

I welcome your suggestions and comments!

Hello @Deeya,

You're right, 60%-80% is generally considered pretty good. Are you losing the majority of reads at the chimeric filtering step? Instead of adjusting the --p-min-fold-parent-over-abundance parameter, have you tried adjusting the trim and truncate parameters? These can significantly alter your % retained without tweaking the algorithm in a way that might light true positive chimeras through.

1 Like

Hi Colin!
Yes! I am losing those reads in the chimera filtration step
I have used the following command
qiime dada2 denoise-paired
--i-demultiplexed-seqs NGS1mod-paired-end-demux-trimmedprimers.qza
--p-trunc-len-f 260
--p-trunc-len-r 220
--o-representative-sequences NGS1modtrimmed-rep-seqs-dada2.qza
--o-table NGS1modtrimmed-table-dada2.qza
--o-denoising-stats NGS1modtrimmed-stats-dada2.qza
And here is the qzv showing the the read quality after trimming of primers
NGS1mod-paired-end-demux-trimmedprimers.qzv (319.8 KB)
Are my truncating parameters too strict?
I might give a try of classifying the sequences taxonomically with these 60-80% non-chimeric sequences and see the result, then decide accordingly. What do you think?

Hello @Deeya,

If you're losing all of the lost reads at the chimera filtering step then adjusting trim and trunc won't make a difference.

I might give a try of classifying the sequences taxonomically with these 60-80% non-chimeric sequences and see the result, then decide accordingly. What do you think?

Yep you can always come back and change things.

1 Like