Relative performance of cluster algorithms and validation indices in maize genome-wide structure patterns

A number of clustering algorithms are available to depict population genetic structure (PGS) with genomic data; however, there is no consensus on which methods the best performing ones. We conducted a simulation study three PGS scenarios subpopulations k = 2, 5 and 10, recreating several maize genomes as model to: (1) compare well-known methods: UPGMA, k-means and, Bayesian method (BM); (2) asses four internal validation indices: CH, Connectivity, Dunn Silhouette, determine reliable groups defining PGS; (3) estimate misclassification rate for each index. Moreover, publicly dataset was used illustrate outcomes our simulation. BM classify individuals in all tested scenarios, without assignment errors. Conversely, UPGMA highest rate. In 10 subpopulations, CH Connectivity indices had maximum underestimation group cluster algorithms. Silhouette showed performance BM. Nevertheless, since measures degree confidence assignment, probability membership, these results should be considered caution. this we found that efficient both simulated real datasets. This offers robust alternative unveil existing PGS, thereby facilitating studies breeding strategies programs. present findings may have implications other crop species.