Strengths & Limitations of GBS

Genotyping by sequencing (GBS) is a reduced representation, next generation sequencing approach that provides a robust and cost-effective means to genotype large numbers of individuals at high density, by targeting sequence adjacent to restriction enzyme cut sites. The robustness and cost-effectiveness of the GBS protocol derives from its relative simplicity. To date, we have genotyped more than 30,000 maize samples via GBS. We have built our own bioinformatics pipeline, part of the TASSEL standalone, for the analysis of this large data set, a key feature of which is its scalability. The main limitation of GBS is the large amount of missing data. To call heterozygous segments, correct errors, and facilitate imputation in biparental populations of recombinant inbred lines (RILs), we have implemented a hidden Markov model (HMM). In contrast, for imputation of unrelated inbred lines, we instead employ a nearest neighbor, fixed window approach. The limited pool of founders contributing to elite US inbreds facilitates their accurate imputation. Furthermore, the bottleneck experienced by maize during its domestication also placed a limit on the number of founding haplotypes. Our current results using GBS data for GWAS and NAM-GWAS will be discussed.