Genome mock to predict single-crosses

Genomic prediction, based on molecular markers, enables speed breeding schemes and increases the response to selection. Even though there are several genotyping platforms for obtaining single nucleotide polymorphism (SNP) markers, lacking comparative information on how these platforms affect hybrid prediction or the inclusion of non-additive effects. Moreover, SNP discovery techniques are commonly based on a unique reference genome, which can introduce an ascertainment bias when tested germplasms are distant from reference. We employed a tropical maize single-crosses panel and genomic data from two genotyping platforms: array and genotyping-by-sequencing, both based on the B73 genome (temperate). Also, we used a pipeline to build a mock reference genome for SNP discovery aiming to capture unique SNP markers within the tropical maize population, independent from an external reference genome. Our results indicate that mock reference genomes deliver reliable estimates for genetic diversity and population structure assessment. Furthermore, genomic prediction estimates were comparable to standard approaches, especially when considering additive effects or simple traits. However, mock genomes were slightly worse to predict complex traits and estimate dominance effects, but still with similar GBS performance using B73 as the reference genome. Nevertheless, the SNP-array methods achieved the best predictive ability and reliability to estimate variance components. Finally, the mock genomes can be a worthy alternative to perform genetic diversity and genomic selection studies, especially for those species where the reference genome is not available.