2021_Saag_EastEuropean - SOP002 - EstCWC has broken genotype data

[nevrome]

Something is wrong with this specific individual, possibly the whole package.

trident forge stops with encountered illegal genotype Hom-Ref with Ref-Allele missing if this individual is included.

Any ideas, @AyGhal and @stschiff?

[AyGhal]

I don't know if it is only one individual, but the convertf issue I mentioned in our meeting was:
1 rs3094315 0.02013 752566 3 1
1 rs12124819 0.020242 776546 0 1
e.g. from this package, using convertf messes up Ref/Alt alleles. Which can also causes problem when using PLINK.
I am redoing this package, but maybe I should check all of them.

[nevrome]

Aha! So this is linked to the issue you mentioned. Would be great if this could be fixed.

Don't worry about the rest of the packages though - we will have to automate this validation anyway (see #6)

[stschiff]

Aha OK, thank you @AyGhal, that helps a lot. I will check whether my current handling of this is too strict. Perhaps I make a tacit assumption that the first allele would always be reference (didn't think so, though).

[stschiff]

OK, I checked. I'm definitely not assuming anything. This is really a hard error because apparently in this genotype data there is a Hom-Ref genotype at a position where the first allele is missing (0). Of course, one way to deal with this without throwing an error is to simply mark the genotype in such a case as missing... but I'm worried that I'm missing something here...

[nevrome]

You can reproduce the error with this code:

mkdir miniposeidon
trident fetch -d miniposeidon -f "*2020_Cassidy_IrishDynastic*,*2021_Saag_EastEuropean*"
trident forge -d miniposeidon -f "<SOP002>,<GNM1007>" -n testpac -o testpac

Error in the genotype data: "encountered illegal genotype Hom-Ref with Ref-Allele missing"

[stschiff]

I just pushed an update to poseidon-hs (bumped version to 0.14.1), which should at least now give more helpful error messages also indicating which SNP caused the error. If you have time to explore this further, feel free, otherwise I'll have time tonight to check myself.

[stschiff]

@AyGhal it's clear now that the data for this package is broken. I doubt that any of our main tools caused this, and I blame simply the combination of conversion steps that were necessary for creating this dataset. Could you generate a new dataset either using trident genoconvert or even directly use the new Plink output option of pileupcaller?

[AyGhal]

Okay, but is the second one also broken?

[stschiff]

What do you mean by "second one"?

[AyGhal]

The one I prepared on Friday, I will share the link.

[stschiff]

OK, I think that works now!

[stschiff]

OK, so I've just tried to forge your new package under /projects1/paleorider/data_published/2021_Saag_EastEurope/genotyping/2021_Saag and that seems to work.

[stschiff]

@AyGhal could you make sure that this new package is correctly in our local SDAG poseidon repo? And also make any necessary updates to the public repo (pull request...). Also, please update the package version. When all that is done I'll replace the version on the server. Thanks!

[nevrome]

I guess that is solved now. 👍