DNA analysis has a very important and scientifically acknowledged role to play in forensics and medicine and law (for example) but the analysis provided by most commercial organisations selling to the general public and claiming deep genealogical insights is little more than modern day snake oil - bunkum (at least to date). See for example DNA ancestry tests branded 'meaningless' and Genetic Ancestry Testing.
Many companies now offer to tell you about your ancestors from a DNA test. Adverts for these tests can give the impression that your results are unique and that the tests will tell you about your specific personal history, but the very same history that you receive could equally be given to thousands of other people. Conversely, the results from your DNA tests could be matched with all sorts of different stories to the one you are given: you cannot look at DNA and read it like a book or a map of a journey. Sense About Genetic Ancestry Testing.
I have no doubt many will disagree (such tests have already been taken by an estimated over 12 million people) and I do not dispute that useful, reliable results have often been achieved that could not have been in any other practical way. Amongst such may be included Christopher Franklin† that, though not for the purposes of genealogical research directly, was a familial comparison.
The dna tag is at present our fourth most popular tag and I believe the matter should be on topic here. However the Stack Exchange family of sites has an excellent reputation and quite rightly strives for high standards. The connection gives much credibility to what many scientists say are as meaningful as a horoscope.
Without seeking to make the subject off topic I feel we have an obligation at least to provide information that warns users of some of the dangers. For example the ones expressed by Scientific American in their article 23andMe Is Terrifying, but Not for the Reasons the FDA Thinks.
Maybe this meta post could be considered adequate for such purposes but, given what seem limited options within the SE format, would a small caveat in tag wikis be appropriate (plural because there are several, see the bullet points in the tag wiki for DNA, and I do not seek to single out 23andMe)?
* Title borrowed from Live Science.
† The article is dated 2010 and since then the jury recommended that Lonnie Franklin Jr. should be sentenced to death.