Randomising plate before sending to dart

[Naomi Brunjes]

Hi there,

Thought I’d reach out via this to see if I can get a response. 

I previously participated the 'Pop-gen in R' course last year and found it super helpful! I’m currently getting ready to send some genetic material to dart and can’t seem to find the part in the course talking about randomising samples. I remember briefly talk about randomising the individuals and locations into the genetic plate before sending away rather than location a, followed by location b, etc, etc. Just wondering if there was a specific layout we should follow. Would you be able to share your thoughts on a good plate layout?

Regards,

Naomi Brunjes

[Peter Unmack]

Samples in the first two or three rows are used as technical replicates,
those first rows should represent the broader diversity present in your
samples. After that it doesn't matter what order they are in.

You can also specify to dart which samples to use as technical
replicates irrespective of where they are on the plate, but any
deviation from their standard approach could easily get missed.

Cheers
Peter

> --
> You received this message because you are subscribed to the Google
> Groups "dartR" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to dartr+un...@googlegroups.com
> <mailto:dartr+un...@googlegroups.com>.
> To view this discussion visit https://groups.google.com/d/msgid/dartr/
> e80cddfd-7091-4438-b6da-0d24220d7b30n%40googlegroups.com <https://
> groups.google.com/d/msgid/dartr/e80cddfd-7091-4438-
> b6da-0d24220d7b30n%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Renee Catullo]

Hi Naomi,

Peter is right - it’s the first two or three rows in each plate that matter most if sending to Diversity Arrays, and it should represent the breadth of diversity.

There isn’t a simple equation you can use, unfortunately. This is because the amount of diversity you are sending varies from others. If you have different species, then equal representation of both species would be best. If you’re just including a few individuals of a second species as an outgroup, then put those individuals up front. If you are looking across a widespread species, you would try to balance across the range. If you don’t know this yet because that’s the question you’re trying to answer, then representing your suspected distinct groups equally is the best option.

And remember that the more divergent your groups are in a single SNP calling exercise, the fewer SNPs you’ll get. This is true for all methods of ddRAD/dartSeq because variable sites become polyallelic and/or the enzyme binding size evolve/fail and you get reduced call rate for loci across groups. You can also get weird biases in your data. This is just an issue around who you include in each set of SNPs, not an issue with the raw sequence data.

When this happens you can improve your power/SNP number by calling SNPs on individuals specific to your question. This can be done by either taking Diversity Arrays up on their offer for more than one set of SNPs from your order, or calling SNPs yourself using Stacks or similar from the raw data. You don’t need to do different orders (the raw data is good), you just need to remember that one set of SNPs can’t answer all questions because of the biases inherent in biallelic data.

Cheers,

Renee

> To unsubscribe from this group and stop receiving emails from it, send an email to dartr+un...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/dartr/fd14d870-5b58-41d4-a123-993d90dc42fc%40unmack.net.

[Naomi Brunjes]

Thanks so much for the responses!!

[nanisrobledo]

Hi Naomi,

Here the video of the excellent talk that Renee gave during the Kioloa workshop about this topic: https://www.youtube.com/watch?v=g1--meLEu0Y

The recordings of all the talks are in the dartR youtube channel.

Best,

Diana