A comparison of polymorphism information content and mean of transformed kinships as criteria for selecting informative subsets of barley (Hordeum vulgare L. s. l.) from the USDA Barley Core Collection

摘要

Recent advances in genetic technologies have given researchers the ability to characterize genetic marker data for large germplasm collections. While some studies are able to capitalize on entire germplasm collections, others, especially those that focus on traits that are difficult to phenotype, instead focus on a subset of the collection. Typically, subsets are selected using phenotypic or geographic data. One major hurdle in identifying favorable subsets is selecting a criterion that can be used to quantify the value of a subset. This study compares two such criteria, polymorphism information content, and a new criterion based on kinship matrices, which will be called the mean of transformed kinships. These criteria were explored in terms of their ability to select subsets that are favorable for genome wide association studies, and in their ability to select subsets that contain a high number of rare phenotypes. Using phenotypic and genotypic data that has been amassed from the USDA Barley Core Collection, evidence was found to support the hypotheses that subsets based on the mean of transformed kinships were well-suited to select subsets intended for genome-wide association studies, but the same was not found for polymorphism information content. Inversely, evidence was found to support the hypothesis that subsets based on polymorphism information content were well-suited to select subsets intended for rare-phenotype discovery, but the same was not found for subsets selected using the mean of transformed kinships criterion. Tools to select subsets using these two criteria have been released in the R package "GeneticSubsetter.''