Looking for population structure. Molecular data mining

PEÑA MALAVERA ANDREA; FERNANDEZ, ELMER ANDRES; BALZARINI MÓNICA

Bogotá

Simposio; Simposio de Estadística; 2011

Universidad Nacional de Colombia

In biological studies identify the underlying genetic structure of a set of individuals is of importance. Several statistical methods for population structure inference from molecular marker data are currently used. We evaluated and compare the following algorithms: hierarchical clustering algorithms, non hierarchical cluster, self-organizing maps, a Bayesian method implemented in the software STRUCTURE and eigen-analysis as tools for looking population structure. We showed that the Bayesian and UPGMA algorithms performed better in most of the evaluated. All the methods performed poorly, when the level of genetic separability among groups is low.