Authors: Nelson Nazzicari Filippo Biscarini Paolo Cozzi E Charles Brummer Paolo Annicchiarico
Publish Date: 2016/05/25
Volume: 36, Issue: 6, Pages: 69-
Abstract
Genotypingbysequencing GBS is a rapid and costeffective genomewide genotyping technique applicable whether a reference genome is available or not Due to the costcoverage tradeoff however GBS typically produces large amounts of missing marker genotypes whose imputation becomes therefore both challenging and critical for later analyses In this work the performance of four general imputation methods Knearest neighbors Random Forest singular value decomposition and mean value and two genotypespecific methods “Beagle” and FILLIN was measured on GBS data from alfalfa Medicago sativa L autotetraploid heterozygous without reference genome and rice Oryza sativa L diploid 100 homozygous with reference genome Alfalfa SNP were aligned on the genome of the closely related species Medicago truncatula L Benchmarks consisted in progressive data filtering for marker call rate up to 70 and increasing proportions up to 20 of known genotypes masked for imputation The relative performance was measured as the total proportion of correctly imputed genotypes globally and within each genotype class two homozygotes in rice two homozygotes and one heterozygote in alfalfa We found that imputation accuracy was robust to increasing missing rates and consistently higher in rice than in alfalfa Accuracy was as high as 90–100 for the major most frequent homozygous genotype but dropped to 80–90 rice and below 30 alfalfa in the minor homozygous genotype Beagle was the best performing method both accuracy and timewise in rice In alfalfa KNNI and RFI gave the highest accuracies but KNNI was much fasterThe rice data used in this research paper were produced within the framework of the Italian national project “RISINNOVA” Grant No 2010–2369 financially supported by the AGER Foundation The creation of the alfalfa data sets was funded by the projects Genomic selection in alfalfa GENALFA funded by the Italian Ministry of Foreign Affairs and International Cooperation in the framework of the ItalyUSA scientific cooperation program the Italian share of the FP7ArimNet project Resilient water and energyefficient forage and feed crops for Mediterranean agricultural systems REFORMA funded by the Italian Ministry of Agricultural and Forestry Policies
Keywords: