The value of mean length after dada2 is too different from the expected value

[colinbrislawn (Colin J Brislawn)]

Hello again @LiyingXie,

@LiyingXie

On the contrary, the number of sequences passed is lower, so I lose too many sequences.

Just as I had feared. Because the region overlapping is low quality, a longer overlap does not cause more reads to join...

@LiyingXie

I found that the passing rate of the sequence is very high, which is very good.

OK, using only the forward read fixed the problems with quality and joining!

@LiyingXie

Moreover, OTU has the same error as before, that is, an OTU only appears in one sample, and the frequency is very low.

That is a problem, but luckily I have found the cause

It looks like your reads start with some random bases, then an adapter, then the true read. I'm willing to bet that these 6 bp are barcodes, causing your ASVs to seperate by sample.

(This paper lists actcctacgggaggcagcag is a 16S primer.)

If you trim off the barcode and adapter, your ASVs should appear across samples!

Try running with --p-trim-left 26 and see how it goes!

---

@LiyingXie

• If we give up the low quality of the two terminal data, is the subsequent diversity analysis still of biological significance? Will abandoned data have a great impact on the overall analysis?

Yes. Removing low quality data should make your diversity analysis more accurate!

@LiyingXie

Could it be that there is something wrong with my rawdata, similar to some abiotic sequences?

Those 6 starting basepairs and adapters look pretty abiotic to me!

@LiyingXie

• Is the similarity of OTU clustering 97% in dada2?

Even better: the ASVs created by DADA2 could be 100% unique, with as little as 1 bp difference between them! The DADA2 paper explains this more.

Keep in touch,
Colin