Genome–wide association study and genomic prediction of growth traits in bighead catfish (Clarias macrocephalus Günther, 1864)

Abstract

Next-generation sequencing (NGS) enables cost-effective exploration of genome-wide single nucleotide polymorphisms (SNPs). This new technology has made enormous contributions in various fields including genetic improvement. This study was conducted to identify SNPs that associated with growth traits (body weight, standard length, and total length) of bighead catfish (Clarias macrocephalus Günther, 1864) using genome-wide association studies (GWAS), and to estimate genetic parameters of these growth traits using genomic best linear unbiased prediction (GBLUP). Individual DNA samples from 991 eight-month-old bighead catfish across 74 full-sibs and 31 half-sibs were collected and subjected to next-generation sequencing using the DArTSeq platform. In return, 9530 SNPs were obtained and used for analysis together with the recorded phenotypic data. The analysis was performed using weighted genomic best linear unbiased prediction (wGBLUP), facilitated by the BLUPF90 family of programs, and identified a set of 19 markers associated with all growth traits; the proportion of explained variance ranged from 1.02 to 6.59%. Four SNP sequences were annotated to Egl nine 2-like gene (egln2), zinc finger FYVE-type containing 21 gene (zfyve21), junctophilin 3a (jph3a), and a hypothetical protein. Furthermore, GBLUP was used to estimate GEBV and heritability of the growth traits. PBLUP was also performed using pedigrees obtained from SNP-based parentage information to estimate PEBV and heritability. The results revealed that GBLUP improved accuracy of EBV estimates over PBLUP. Heritability of growth traits estimated by GBLUP ranged between 0.29 ± 0.05 (for standard length) and 0.55 ± 0.06 (for body weight), with accuracy ranging from 0.672 to 0.724 and prediction bias close to one. When compared to GBLUP, heritability estimated by PBLUP was lower, and with lower accuracy, while bias was lower only for standard length. In conclusion, the growth-associated SNPs identified herein explained only a small portion of genetic variance, and thus using these SNPs for marker-assisted selection is not promising. Instead, genomic selection based on GEBV is a better approach, provided that genotyping cost can be optimized. The knowledge gained from this study is beneficial to genetic improvement of bighead catfish and other related species.