Mathematicians like to say: "Don't believe any statistics you haven't made yourself". But it is a fact, that especially in very extensive data sets, almost all "desired" results can be found.
On the other hand, this is not very helpful in the long run, especially when the goal is to make reliable statements. That's why at lifespin we never rely on a result obtained by using just one mathematical method.
By default, we test by mathematical methods that calculate how likely it is that the result is a random product of the numbers at hand. Further we always secure the results multicenter. I.e. only if material from two or more different and independent sources delivers the same result, we speak of a "reference in the data".
However, we only finally accept a result when mathematics and biology come to the same conclusion. That is, when the abnormality found is also biologically meaningful and explainable, e.g. when the corresponding enzymes and metabolic cascades are understood, described and comprehensible.