Problems with defining denoising parameters to improve data results

Hello,

I'm having issues in choosing the denoising parameters. I have testes many possibilities, but after denoising I get about 60% of of input passed filter and about 15-30% of input non-chimeric. I'm not sure if is the sequencing quality or else. Could anyone help me, please?

So, this is a 341F-805R 16S rRNA sequencing of samples from anaerobic digestors. We did a Illumina 2x250bp (50,000 to 100,000 reads per sample).

Here there is the quality plot.

Here there is the sumamry.

Here there is the last command I've tried:
qiime dada2 denoise-paired --i-demultiplexed-seqs trimmed.qza --p-trunc-len-f 245 --p-trunc-len-r 240 --p-n-threads 0 --o-table dada2-table-12.qza --o-representative-sequences dada2-repseqs-12.qza --o-denoising-stats dada2-denoisingstats-12.qza

Here there is the denoisng stats.

I appreciate the help very much.
Cheers.

1 Like

Hi @carolinnerdc

Your current parameters for trimming your forward and reverse reads might be too generous, in that you’re dipping into parts of the reads which significantly drop in quality.

I would try —p-trunc-len-f 210 and —p-trunc-len-r 220.

See if that improves your denoise stats. Best of luck!

Thank you @Mike_Stevenson

I tried your suggestion and that is the denoising stats I got. I think it is not right :upside_down_face:
Do you have an idea of what could be happening?

Hi @carolinnerdc

It looks like you were able to get more reads passing the initial filter step but the problem now lies in the number of reads merging (or not merging as it were).

DADA2 requires 12bp overlap in order to merge reads correctly. I would try —p-trunc-len-f 200 and leave the reverse at 220. This should allow for merging to occur and increase the number of non-chimeric reads.

Good luck!

1 Like

I agree

44% merged sounds bad...
16% non-chimeric sounds worse.

9k reads per sample sounds pretty good!

Given the poor run quality, this is a surprisingly good result.
What does this look like on your full run?

2 Likes

I did the full run with Mike's last suggestion.
For the denoising stats, it didn't changed much. But, I could get the results I was expecting. I read somewhere, I dont remember which forum, that I could get stats like that if I had a small diversity in my samples, which is kind of expected by me. They come from a semi-continuos anaerobic digestor fed with sugarcane bagasse that have been running since last December in a 55º C temperature. Even though the samples come from very diverse environment, like soil, by now the selective pressure might have act already. The groups I could get are mainly Clostridium and Archaea, and that makes a lot of sense to us.

Do you guys think that could be it? :thinking:

1 Like

Maybe?

Can you post the denoising stats from the full run?

Those questions seem to be related to alpha diversity, which you can run after DADA2!

1 Like

Most of the results I am getting are with this values.

1 Like

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.