quality score plot DADA2

[murheisa]

Hello, I'm currently working on some 16S sequences of whale microbiomes, they were sequenced with Illumina, V4 region, and I´ve been familiar only with V3V4, so I don't know if this quality and at the beginning of the sequence is normal. In the tutorial you mention that in V4 region there are sufficient overlap, so you can guide the trimming parameters only with quality, but these samples seem to me little odd. Can I trim 30 pb at the beginning ?

[benjjneb]

The red line is showing the number of reads that extend at least that far. The lower quality at the start is due to a large fraction (80%?) of reads that are only ~30nts long in your data.

How has this data been preprocessed and do you have any ideas on why there are a very large number of reads that are both bad quality and only 30 nts long?

But the solution to this would probably not be trimming the starting reads, but removing these short reads.

[murheisa]

The sequences are supposed to be raw reads, and all samples have this same issue, I checked with fastqc and all of these short sequences are only NNNNNNNNNNN. The sequencing company used these raw reads with qiime with a previous removal of primer with fastx and quality values with less than 20 as well as sequences less than 130 bases in length were removed and merged with FLASH script and then used in qiime. These were 2x300 bp.

I wonder, if I remove only these 30 pb reads with NNNNNN in trimgalore, the sequences that are kept, could I used them in dada2, and then with filter and trim remove primers and low quality as usual?

[benjjneb]

I wonder, if I remove only these 30 pb reads with NNNNNN in trimgalore, the sequences that are kept, could I used them in dada2, and then with filter and trim remove primers and low quality as usual?

Yes, you can work just with the remaining reads. And short all-N reads will be removed by filterAndTrim(..., minLen=100) as well.