Have questions? Visit https://www.reddit.com/r/SNPedia

Talk:Gs144/criteria

From SNPedia

What exactly is the logic that is being used here in determining sex? I’m seeing some people who have sex miscalled (probably?) in their promethease reports, in particular, women being called as male. I note that there is a requirement of at least 2 of the list of SNPs which appear to be Y related.

What would be the expected number of SNPs present in a biological male, and what would be the expected number of SNPs present in a biological female? (Guessing the latter is 0, and the 2 is set so that 1 miscall won’t screw up the algorithm, but depending on the expected number of SNPs in males, perhaps the threshold could be raised without loss of sensitivity?)

Allow me to guess that the problems you've seen are specific to data from Ancestry.com. Data from Ancestry.com is the only major vendor which consistently shows bad calls on the sex chromosomes. Most of the calls are correct, but it's quite common to have heterozygous Y snps (outside the pseudoautosomal), and some Y calls for females. The problem is bad enough that we've had to do special handling for that platform. So for Ancestry.com, this criteria is ignored and instead we use various heuristics across ALL snps in the data, not just the subset listed here. At present, one of the criteria the Ancestry specific code depends on is seeing a minimum of 100 Y chrom snps to confidently call male. We consistently see some true males who's data does not meet this standard. I've not yet decided whether to lower the bar, or make a new genoset that shows for these ambiguous cases. --- cariaso 11:09, 8 May 2018 (UTC)