Inconsistent primer trimming sequence in amptk_mock*.fa about amptk

[nextgenusfs]

I will have to look back and get confirmation, but my recollection was that these sequences were from Sanger sequencing of the actual clones -- where we might have used a different forward sequencing primer and/or the Sanger data wasn't good. Surely that is where the Y came from (ie Sanger data was inconclusive at that position).

[peterjc]

Yes, the ambiguity is undoubtedly from Sanger capillary sequencing - I was puzzled that it was Y in one file, but the more vague N in another.

My main query was about the trimming being different between the three files under https://github.com/nextgenusfs/amptk/tree/master/amptk/DB - but I may not have understood why there are three files?

$ grep -c "^>" amptk_mock*.fa
amptk_mock1.fa:20
amptk_mock2.fa:26
amptk_mock3.fa:23

Given they have differing numbers of sequences, my assumption was three different mock communities. Quoting the paper,

To construct the biological mock community (BioMock) we selected 26 identified fungal
cultures (Table S1) ...

and:

The BioMock-standards consist of an equimolar mixture of 26 PCR
products thereby removing the PCR bias from mixed DNA samples, while the BioMock
communities consist of an equimolar mixture of 23 single-copy plasmids.

That suggests amptk_mock2.fa is the BioMock-standards (equimolar mixture of 26 PCR products), and amptk_mock3.fa is the BioMock communities (equimolar mixture of 23 single-copy plasmids), leaving amptk_mock1.fa unexplained - perhaps a precursor dataset?

[nextgenusfs]

Hi @peterjc. Yes I think you have it figured out. I had used this to just keep track of what we were running when were were doing these experiments and writing this pipeline. amptk_mock1.fa was our first control sample we ran on Ion Torrent and once we saw results we added a few more species to capture some more variability and make sure we understood what was going on (It actually pre-dates the data in the publication, so at this point it probably shouldn't be in the repo other than for legacy/historical reasons).

So the numerical suffix is the iterations of the different "biological mock" communities that we ran. We had included a few clones in biomock2 that consistently did not sequence well on Ion Torrent for different reasons (mind you this is before we developed the synthetic mock) we then attempted to pare down to the 23 that had enough sequence divergence and give us some confidence in the data analysis pipelines.

The more times we ran this the more we got the idea that the biological sequences are problematic due to index-bleed, hence synthetic mocks as the final solution. I still recommend people run the synthetic mock along with a biological mock community that is representative of species/ITS sequences they are looking to differentiate. The other practical advice when using Illumina is just never re-use any of the index/barcodes and then "index-bleed" is actually quite low on MiSeq platform. I don't think many folks are using Ion Torrent anymore, so practical advice would be to buy the MiSeq instead of the Ion Torrent :).

[peterjc]

That makes sense - as much as we try to document as we go along, there will always be hidden assumptions it takes fresh eyes to see.

I found your work looking for other people doing metabarcoding with synthetic controls. Our paper plans are all delayed though lockdown, but many of our methodical conclusions agree (we're only using the MiSeq). Expect a citation to follow...

So, as to this issue (#89), amptk_mock1.fa was used in pilot data before the work described on the paper, so we shouldn't read too much into it differing from amptk_mock2.fa and amptk_mock3.fa. I will close this issue now, thank you.

[nextgenusfs]

Cool, looking forward to the publication! It took me awhile to remember the different mocks that we tried -- most of this work/sequencing was done in 2014-2016 time frame so hard to recall all the details!