DADA2 analysis for two different platforms MISEQ and Novaseq -Merging of sequence table

[Jayakanthankk]

I have sequenced V3V4 region 50 samples in each by two platforms MISEQ and Novaseq and two different formats 2x250 and 2x300 bp. Is possible to merge sequences from both runs and construct sequence tables, followed by assigning taxonomy. Basically all samples of same cohort. How to analyses and what codes required for this.

[benjjneb]

Yes, as long as the primer sets you are using are the same you can simply merge ASVs across those sequencing types.

[Jayakanthankk]

We have sequenced V3-V4 region in two platforms: Miseq 2x250 bp (batch1) and Novaseq 2x300 bp (batch2), not samples and but same cohort. I have trimmed fastq files of Novaseq to 250 bp. I have pooled both data and analyzed it at the same time. I have trim 17 and 21 base pairs to remove primers that are present in samples. The problem here is that each ASV count was present in either of the data sequences. For example, ASV1 is present only in Miseq reads but not in Novaseq, and for ASV2, it is vice versa. Even though primers were the same, ASV generate behaves differently and is platform-specific. What would be the reason for this and how to overcome this issue?

[benjjneb]

Sorry for slow response.

This complete non-overlap between the two batches indicates that something is not matching up between the way you are trimming the data down to a common ASV (i.e. common start/end point). There are a couple technical reasons this could happen, e.g. primers were sequencd in one batch and not the other, or there is some padding bases on one of the runs but not the other.

A way to investigate would be to find two corresponding ASVs from each batch, e.g. the most abundant ASVs that have the same taxonomic classification, and align them. Their should be a findable pair like this that perfectly matches except for some overhang of one sequence versus the other. That will give a clue as to where the trimming/truncation issue lies.