Finding genetic signatures of local adaptation is of great interest for many population genetic studies. Common approaches to sorting selective loci from their genomic background focus on the extreme values of the fixation index, Fst, across loci. However, the computation of the fixation index becomes challenging when the population is genetically continuous, when predefining subpopulations is a difficult task, and in the presence of admixed individuals in the sample. In this paper, we present a new method to identify loci under selection based on an extension of the Fst statistic to samples with admixed individuals. In our approach, Fst values are computed from the coefficients obtained with ancestry estimation programs. More specifically, we used factor models to estimate Fst, and we compared our neutrality tests with those derived from a principal component analysis approach. The performances of the tests were illustrated using simulated data, by re-analyzing genomic data from European lines of the plant species Arabidopsis thaliana, and by re-analyzing human genomic data from the population reference sample, POPRES.