For denoising problem. Is there a bug in 2018.11 version in denoising?

Hi all,
Is there a bug in 2018.11 version for denoising because the number of filter and denoising is same.
But in 2019.4 version the number of of filter and denoising is different.

I wonder which is correct. Any ideas?

Hi @guojun_wu! I assume by “denoising” you are referring to q2-dada2?

The version of DADA2 used by q2-dada2 2018.11 is different than the version of DADA2 used by q2-dada2 2019.4, which is why you might see slightly different results.

1 Like

Hi @thermokarst Thank you for your answer. Yes it is q2-data2. Perhaps it is not a slightly difference. For our data, the ASV number decreased from 800+(2018.11) to 700+ (2019.4), around 100.
I just wonder what’s the difference between the two version of dada2 and Why? The old one seems it corrects the reads and will not lose reads (# filter = # doeniosing) . The new one it also removes some reads (# filter > # doeniosing)

3 Likes

I suggest you take a look at the dada2 docs, you should be able to find release notes there:

https://benjjneb.github.io/dada2/

Hi @thermokarst. I read the release note https://benjjneb.github.io/dada2/ReleaseNotes_1_12.html. But it can not answer what I found. Do you have clues?

QIIME 2 doesn’t use that release, check these notes, instead: https://benjjneb.github.io/dada2/ReleaseNotes_1_10.html

Generally speaking though, there are differences between 1.8 and 1.10 that subtly change the output. If you have more questions, I encourage you to reach out to the DADA2 team directly!