Expected Genotype Quality and Diploidized Marker Data from Genotyping-by-Sequencing of Urochloa spp. Tetraploids

Abstract

Although genotyping-by-sequencing (GBS) is a well-established marker technology in diploids, the development of best practices for tetraploid species is a topic of current research. We determined the theoretical relationship between read depth and expected genotype quality (EGQ) for tetraploid vs. diploidized genotype calls. If the GBS method has 1% error, then 17 reads are needed to classify tetraploid samples as heterozygous vs. homozygous with 95% accuracy, compared with 63 reads to determine allele dosage. We developed an R script to convert tetraploid GBS data in Variant Call Format (VCF) into diploidized genotype calls and applied it to 267 interspecific hybrids of the tetraploid forage grass Urochloa (syn. Brachiaria). When reads were aligned to a mock reference genome created from GBS data of the U. brizantha cultivar 'Marandu', 25,678 bi-allelic SNPs were discovered, compared to approximately 3000SNPs when aligning to the closest true reference genomes, Setaria viridis and S. italica. Crossvalidation revealed that missing genotypes were imputed by the Random Forest method with a median accuracy of 0.85, regardless of heterozygote frequency. Using the Urochloa spp. hybrids, we illustrated how filtering samples based only on GQ creates genotype bias; a depth threshold with corresponding EGQ equal to the GQ threshold is also needed, regardless of whether genotypes are called using a diploidized or allele dosage model.

Association Mapping Considering Allele Dosage: An Example of Forage Traits in an Interspecific Segmental Allotetraploid Urochloa spp. Panel

“…S2). Setaria italica allowed more alignments for Urochloa spp., which were relatively well distributed across the genome, following what was observed by Matias et al (2019).…”

“…S1). These markers came from centroids in the Um genome reference (Matias et al, 2019) originated from the GBS‐SNP‐CROP approach (Melo et al, 2016). Although there were no candidate genes on these regions, the last part of both sequences showed the same final nucleotide sequence.…”

“…( Si ) (Bennetzen et al, 2012), Sorghum bicolor (L.) Moench. ( Sb ) (DOE‐JGI, 2018b), Oryza sativa L. ( Os ) (Ouyang et al, 2006), Zea mays L. ( Zm ) (Schnable et al, 2009), and the Urochloa mock reference ( Um ) (Matias et al, 2019). Default alignment parameters were used in the software Burrows‐Wheeler Alignment tool (Li and Durbin, 2009), SAMtools (Li et al, 2009; Li, 2011) and Picard (http://broadinstitute.github.io/picard/).…”

“…The average depth was 18 reads per marker by sample. Remaining missing data were imputed using the Random Forest package (Liaw and Wiener, 2002), where all markers with r 2 ≥0.1 were used as predictors, and 300 trees were considered, following the recommendation of Matias et al (2019). Consequently, we selected 26,535 SNPs with diploid and tetraploid dosage configurations.…”

“…For significant SNPs found on the Urochloa mock reference (Matias et al, 2019), a sequence of 50 nucleotide positions from both sides of significant SNPs was selected from the respective reference and compared with the information available in genomic data banks using the Basic Local Alignment Search Tool (BLAST) (Altschul et al, 1990). For the significant SNPs found on the other references (annotated genomes), a window of 1 kb was evaluated to determined candidate genes.…”