You'd be surprised. There is quite a bit of polymorphism within the human species that is very much distinguishable per population. E.g. haplogroup analysis or microsatellite analysis is remarkably accurate in this regard due to a lack of interaction between far flung populations until quite recently in human history. Now, does this imply all the bullshit eugenicists and other racists tend to preach about with race? Hell no, social factors are responsible for most of that variance, but to suggest there would be no biomarker for "race" in its colloquial definition as proxy for population of origin is inaccurate.
This is also why there is a big focus now to seek out underrepresented populations in genetic analysis, because there may be population specific biomarkers that are relevant in disease that you miss if you limit yourself to the handful of widely sequenced homogeneous populations (e.g. there are Utah and Iceland datasets that are popular to use for this).