How is error probability estimated while calculating phred quality scores? I have multi samples in my vcf file and I see many SNP's having a quality score of '999'. Is there any threshold value after which the SNP's are assigned a quality score of '999'.
The limits are probably hard coded into the software - a 999 value is probably unrealistic anyway. If not from mathematical perspective, there may be good reasons for a value to be this but from biological or experimental perspective. The underlying error rates of the experimental protocol or sequencing platform most likely far exceed the reported value that simply relies on the results that are most certainly biased above those values.