Converting Allele probs to genotype probs?

[CJ Harbort]

I have a dataset of DO mice from a publication, where they have only included the calculated allele probabilities (probabilities for the 8 founder line alleles), but I would like to have the genotypes of the mice at some loci of interest. Essentially I need the reverse of genoprob_to_alleleprob(), so I could then feed this into maxmarg(). 

Is there a function for this or another workaround? Besides contacting the authors and hoping they can provide the genoprobs :)

Thanks in advance for any ideas!

CJ

[Karl Broman]

It’s not generally possible; going from 36 probabilities to 8 dosages can’t be reversed.

Perhaps if the values were all 0 with one 1 (so complete information), but I don’t have a function for it.

karl

[CJ Harbort]

I thought that might be the answer. Thanks anyway for taking the time. 

Really impressive work with the package Karl! And amazing effort you put in to all the support. 

Best,

CJ

--
You received this message because you are subscribed to a topic in the Google Groups "R/qtl2 discussion" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/rqtl2-disc/zW2oyPL95Nc/unsubscribe.
To unsubscribe from this group and all its topics, send an email to rqtl2-disc+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/rqtl2-disc/34f8452b-d6d7-4797-a286-455ec27145fen%40googlegroups.com.

[Karl Broman]

Here's a stab at a function if you want to try. It looks for cases where one allele is probability > 1 - epsilon or two alleles both have values > (1-epsilon)/2

    https://gist.github.com/kbroman/63dcb6ec63a007415d450ad188335dee

karl

[CJ Harbort]

Perfect! Thanks a lot

CJ

To view this discussion on the web visit https://groups.google.com/d/msgid/rqtl2-disc/dcf6a1ec-cc83-4e12-bcc2-609261f58968n%40googlegroups.com.