Getting the numbers Plink used for hardcall distances?
Kristen Kelly
Hello,
I can see in the documentation for --hard-call-threshold that by it uses a threshold of a distance of 0.1 from the nearest hardcall to determine whether to set a genotype to missing when converting to hardcalls.
If I understand your distance calculation correctly, it’s based on dose for BOTH alleles, so if there was a C/T SNP and someone had a dosage of C=1.15/T=0.85 the calculation would be 0.5*(abs(1.15 – 1) + abs(0.85 – 1)) = 0.5*0.3 = distance of 0.15. Is that correct?
And if I want to calculate what the distance would have been for certain people on certain SNPs, can I use data exported with --export A to do that? Eg. if the .raw file shows the dose for one allele is 1.40784, does that mean the dose for the other allele is 2 - 1.40784 = 0.59216, and I can then use those two dosages to calculate distance? Or are there ever circumstances where the two dosages may not sum to 2 so I can’t infer what the other dosage was from the dosage shown in the .raw file?
Thank you,
Kristen
Chris Chang
2. Yes, the output from --export A contains all the information used by --hard-call-threshold, and it is applied in the manner you describe.
--
You received this message because you are subscribed to the Google Groups "plink2-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plink2-users...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/plink2-users/SJ0PR03MB583997B618D5A47F2BCC7A259293A%40SJ0PR03MB5839.namprd03.prod.outlook.com.