The trade-off between density marker panels size and predictive ability of genomic prediction for agronomic traits in Coffea canephora

Abstract

Genomic prediction in Coffee breeding has shown good potential in predictive ability (PA), genetic gains and reduction of the selection cycle time. It is known that the cost of genotyping was prohibitive for many species, and their value is associated with the density markers panel used. The use of optimize marker density panel may reduce the genotyping cost and improve the PA. We aimed to evaluate the trade-off between density marker panels size and the PA for eight agronomic traits in Coffea canephora using machine learning algorithms. These approaches were compared with BLASSO method. The used data consisted of 165 genotypes of C. canephora genotyped with 14,387 SNP markers. The plants were phenotyped for vegetative vigor (Vig), rust (Rus) and cercosporiose incidence (Cer), fruit maturation time (Mat), fruit size (FS), plant height (PH), diameter of the canopy projection (DC) and yield (Y). Twelve different density marker panels were used. The common trend observed in the analysis shows an increase of the PA as the number of markers decreases, having a peak when used between 500 and 1,000 markers. Comparing the best and the worse results (full SNP panel density) for each trait, some had an improvement around of 100% (PH: 0.35–0.77; Cer: 0.40–0.84; DC: 0.39–0.82; Rus: 0.39–0.83, Vig: 0.40–0.77), the other showed an improvement more than 340% (Mat: 0.12–0.60; Y: 0.14–0.61; FS: 0.07–0.60). The results of the current study indicate that the reduction of the number of markers can improve the selection of individuals at a lower cost.