With most genetic genealogy DNA companies, they get parent/child and full siblings right. The exception would be if a parent was an identical twin. As the DNA companies test only a fraction of the over 3 billion base pairs in your DNA, once you get past these two relationships, it becomes harder. Here’s 23andMe’s estimate: https://customercare.23andme.com/hc/en-us/articles/212170958-DNA-Relatives-Detecting-Relatives-and-Predicting-Relationships#percent_shared. I disagree with identical twins sharing 100% DNA. It’s close, but some of my earlier posts some identical twins have minor differences between them. Not enough to make a difference with companies like 23andMe that only test a small fraction of DNA.
In some cases, no calls occur occasionally with a sample. Not enough to skew results as companies look at results using quality control tool measures to catch large variations. They also do random pulls for quality control purposes and run those a second time to validate the system works.
I see some people claim DNA tests are inaccurate. When done correctly for close relationships, they are accurate within the limits noted in 23andMe link above. For example, 25% shared DNA can fall into several categories. Age can rule grandparent/grandchild.
For other relationships, not as clear cut. Mom is 23 years younger than her half-brother. Years ago, I was watching a 50 year actress being interviewed. She looked pregnant. In the interview, she announced she was pregnant. She had twins soon after. She had a child from an earlier marriage.
Previous DNA Saturdays posts: https://upsdownsfamilyhistory.wordpress.com/tag/DNA-Saturdays/