Accuracies of genomic predictions for disease resistance of striped catfish to Edwardsiella ictaluri using artificial intelligence algorithms

Abstract

Assessments of genomic prediction accuracies using artificial intelligence (AI) algorithms (i.e.,, machine and deep learning methods) are currently not available or very limited in aquaculture species. The principal aim of this study was to examine the predictive performance of these new methods for disease resistance to Edwardsiella ictaluri in a population of striped catfish Pangasianodon hypophthalmus and to make comparisons with four common methods, i.e.,, pedigree-based best linear unbiased prediction (PBLUP), genomic-based best linear unbiased prediction (GBLUP), single-step GBLUP (ssGBLUP) and a non-linear Bayesian approach (notably BayesR). Our analyses using machine learning (i.e.,, ML-KAML) and deep learning (i.e.,, DL-MLP and DL-CNN) together with the four common methods (PBLUP, GBLUP, ssGBLUP and BayesR) were conducted for two main disease resistance traits (i.e.,, survival status coded as 0 and 1 and survival time, i.e.,, days that the animals were still alive after the challenge test) in a pedigree consisting of 560 individual animals (490 offspring and 70 parents) genotyped for 14,154 Single Nucleotide Polymorphism (SNPs). The results using 6,470 SNPs after quality control showed that machine learning methods outperformed PBLUP, GBLUP and ssGBLUP, with the increases in the prediction accuracies for both traits by 9.1–15.4%. However, the prediction accuracies obtained from machine learning methods were comparable to those estimated using BayesR. Imputation of missing genotypes using AlphaFamImpute increased the prediction accuracies by 5.3–19.2% in all the methods and data used. On the other hand, there were insignificant decreases (0.3–5.6%) in the prediction accuracies for both survival status and survival time when multivariate models were used in comparison to univariate analyses. Interestingly, the genomic prediction accuracies based on only highly significant SNPs (P < 0.00001, 318 - 400 SNPs for survival status and 1,362–1,589 SNPs for survival time) were somewhat lower (0.3 to 15.6%) than those obtained from the whole set of 6,470 SNPs. In most of our analyses, the accuracies of genomic prediction were somewhat higher for survival time than survival status (0/1 data). It is concluded that although there are prospects for the application of genomic selection to increase disease resistance to Edwardsiella ictaluri in striped catfish breeding programs, further evaluation of these methods should be made in independent families/populations when more data are accumulated in future generations to avoid possible biases in the genetic parameters estimates and prediction accuracies for the disease resistant traits studied in this population of striped catfish P. hypophthalmus.

Effects of common full-sib families on accuracy of genomic prediction for tagging weight in striped catfish Pangasianodon hypophthalmus

“…Inclusion of highly significant SNPs in genomic prediction models that included the c 2 did not have noticeable impacts on the prediction accuracy for tagging weight. This is likely due to the limited size of the significant SNPs obtained from genotyping by sequencing (GBS) platform but our observation here is consistent with previous findings for disease resistance traits in the same population of striped catfish P. hypophthalmus (Vu et al, 2021). In studies where the c 2 estimates were not included, Luo et al (2021) also found there were no advantages of preselected SNPs in genomic prediction models using ssGBLUP, WssGBLUP and BayesB for resistance to Edwardsiella tarda that causes acute symptoms with ascites in Japanese flounder (Paralichthys olivaceus).…”

“…To date, studies in aquaculture species performed multi-trait genomic prediction are limited. Results from these studies showed that the accuracies of genomic predictions were not improved for fillet weight and fillet yield in Nile tilapia O. niloticus ( Joshi et al, 2020 ) or for survival status and survival time in striped catfish P. hypophthalmus ( Vu et al, 2021 ), likely because the high heritability of these two traits and their high genetic correlations; hence, adding one trait did not improve the prediction accuracy of the other. In yellowtail kingfish, Nguyen et al (2022) also showed that the benefits of multi- vs. univariate analysis depend on statistical methods used and genomic architecture of traits.…”

“…However, other studies, which used prioritised variants from GWAS, reported there was an improvement in the prediction accuracy by 1.2%–13.3% for growth-related traits under chronic thermal stress in rainbow trout O. mykiss ( Yoshida and Yáñez, 2021 ) or disease resistance traits in whiteleg shrimp L. vannamei , Atlantic salmon S. salar and gilthead sea bream S. aurata ( Luo et al, 2021 ). In addition, the variant (or marker) effects can be weighed to improve the prediction accuracy as demonstrated in our recent study for disease traits ( Vu et al, 2021 ) or for production traits in dairy cattle ( Xiang et al, 2021 ).…”

“…The bivariate model was also performed in AIREMLf90. In addition, we analysed the two-trait model [tag weight and survival time ( Vu et al, 2021 )], using Gibb Sampling method in THRGIBBF1f90 ( Tsuruta and Misztal, 2006 ). In both packages, the bivariate model was the same as Eq.…”

Insight into selective breeding for robustness based on field survival records: New genetic evaluation of survival traits in pacific white shrimp (Penaeus vannamei) breeding line

“…Here we compared two types of survival definitions; binary traits and continuous traits for our genetic evaluation, with both showing similar heritability estimates ( Table 3 ). While this result did not fit our earlier assumption that continuous traits would improve the results for heritability estimates via adding more survival time information to mortality events, similar findings have been reported for genetic analysis of survival data in experimental challenge tests on aquaculture species ( Joshi et al, 2021a ; Vu et al, 2022 ). Binary trait recording however, is much more simple (0, 1) to score than continuous survival data in practice and thus, should be considered more feasible for routine genetic evaluation in aquaculture breeding programs.…”