Instead of duplicates, I would say one should use replicates and more than two. How would you explain/prove that whatever you see as SNP is real and not artifact/error without any comparison with same or similar samples?
Abhijeet Singh , thanks for your answer. We are just started to use Taqman SNP assay in the lab, and after the first round this question came up. We had 45 samples in duplicates (each sample was pipetted to two different wells and measured indenpendently), and we always got the same result. It looked very robust and reproducible to me, thats why this question came up. Thank you very much again for this response.
Hi, I would like to add that when we work with TaqMan for genotyping - always 1 sample / 1 well. When we do repeats (duplicates, triplicates, even 5 wells per 1 sample) - always the same result.
When we work with Sybr Green based melt curve analysis for insertions and deletions, or even use HRM - triplicates.