Entering edit mode
5.5 years ago
Famf
▴
30
I am using chi-square test to search signals of segregation distortion in a large SNP data set genotyped in large population. Into my results there are many SNPs that look more like genotyping errors rather than segregation distortion signals. Eg. parents AA x AB => progeny 97 AA : 3 AB. As result the chisq value is quite high (very low p-value). So, is there is any general criterion to set a threshold in which we can consider a SNP call as genotyping error (false SD)?. I think it is important to mention it is GBS data.
I think you misunderstand what a statistical hypothesis test and a p-value are. They inform you on the likelihood of observing this or more extreme data assuming that the null hypothesis is true (i.e. usually that there is no difference between what's being compared). This says nothing about the underlying cause or even the size of the effect. So your test points to something unexpected, now you need to investigate the cause. Changing the threshold will only change the number of cases you would consider unexpected under the null hypothesis but still won't inform on the cause.