modifying DO allele probabilities

[Fernando Andrade]

Dear Karl,

I am working with the DO but, because of some particularities of our aims I need to change the allele probabilities (8 alleles for each locus for each individual) into a "carrier" probability. So, for example, if an individual is AB for a locus i guess it would have 1s for these two alleles but 0s otherwise. If the individual is AA it has a 1 for the A allele and 0 otherwise.

So, my question is, is it possible to do this on qtl2? I though about just modifying the allele probability data, but the probabilities would be different between heterozygotes (it would sum 2) and homozygotes(it would sum 1).

Thank you very much for your attention

Fernando

[Karl Broman]

I'm sure it's possible. The probabilities are a list of 3d arrays, individuals x genotypes x positions

karl

[Dan Gatti]

It sounds like you want to round 2 times the 36-state genoprobs. This will give you 0, 1 or 2 for most markers. But you will find markers where you’ll find three 1s instead of two, or all 0s.

 

Would that work for you?

 

Dan

--
You received this message because you are subscribed to the Google Groups "R/qtl2 discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rqtl2-disc+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/rqtl2-disc/c584c8e4-57fb-4180-8ce9-5c0432c4c984n%40googlegroups.com.

---

The information in this email, including attachments, may be confidential and is intended solely for the addressee(s). If you believe you received this email by mistake, please notify the sender by return email as soon as possible.

[Fernando Andrade]

Yes, I think it is something like that. as the data is smaller, I was actually working on the allele probabilities (8 states only per locus) and I though about modifying them, but I am unsure as whether that would impact the analysis scripts, in case that it is assumed that the sum of each row should be ==1. Would you guys say it would be feasible to do it? Am I going to "break" any port of the code?

Thanks for your attention

[Dan Gatti]

So, the rounding operation needs values multiplied by 2 to round correctly. Then you could divide the results by 2 and get back to a 0 to 1 range.

 

But, to clarify, I would NOT use that for linkage mapping. You should use the 8-state allele probabilities for mapping to reflect the uncertainty in the gentyping.

 

Dan

To view this discussion visit https://groups.google.com/d/msgid/rqtl2-disc/6659052a-8f68-4ee3-97ce-de38d45be359n%40googlegroups.com.