The quest for genes representing genetic relationships of strains or individuals within populations and their evolutionary history is acquiring a novel dimension of complexity with the advancement of next-generation sequencing (NGS) technologies. In fact, sequencing an entire genome uncovers genetic variation in coding and non-coding regions and offers the possibility of studying Saccharomyces cerevisiae populations at the strain level. Nevertheless, the disadvantageous cost-benefit ratio (the amount of details disclosed by NGS against the time-expensive and expertise-demanding data assembly process) still precludes the application of these techniques to the routinely assignment of yeast strains, making the selection of the most reliable molecular markers greatly desirable. In this work we propose an original computational approach to discover genes that can be used as a descriptor of the population structure. We found 13 genes whose variability can be used to recapitulate the phylogeny obtained from genome-wide sequences. The same approach that we prove to be successful in yeasts can be generalized to any other population of individuals given the availability of high-quality genomic sequences and of a clear population structure to be targeted.