Genetic Map Construction with Incomplete Marker Information in Maize.

Next generation sequencing technologies provide high-density genetic marker information at a low cost per data point. This increase in marker density allows for the ability to more accurately assess regions of recombination and thus, allow for more precise identification of quantitative trait loci (QTL) underlying traits of interest. A genotyping-by-sequencing (GBS) approach was implemented on recombinant-inbred populations from the crosses Oh43 x W64a, Ny821 x H99 and intermated B73 x Mo17 (IBM) to produce a dense genetic map. A sliding window approach  was implemented to impute missing markers, remove genotyping errors and reduce marker redundancy. The approach consists of identifying areas of recombination and the development of bin markers that flank each recombination breakpoint. The maps obtained were compared with existing Simple Sequence Repeat, Single Nucleotide Polymorphism, and insertion/deletion marker data on these populations to validate the approach. The ability to accurately impute missing data is directly proportional to the level of missing information in the dataset. However, despite the high level of missing data, this approach was able to impute individuals up to 98% accurately without the use of any additional information. The resulting maps consist of 5,320, 5,683 and 8,224 bin markers with an average spacing of 0.57, 0.56 and 0.74 cM for the Ny821 x H99, Oh43 x W64a, and B73 x Mo17 populations, respectively. A reduction in the size of 1.5 LOD support intervals was observed for QTL identified in the IBM population, ranging from 0.76 Mb to 7.23 Mb when compared to previously published genetic maps. The flexibility of this method allows for the incorporation of additional marker information prior to the sliding window approach to further increase imputation accuracy and thus provide concise and accurate marker information for use in mapping studies.