Using encrypted genotypes and phenotypes for collaborative genomic analyses to maintain data confidentiality

To adhere to and capitalize on the benefits of the FAIR (findable, accessible, interoperable, and reusable) principles in agricultural genome-to-phenome studies, it is crucial to address privacy and intellectual property issues that prevent sharing and reuse of data in research and industry. Direct sharing of genotype and phenotype data is often prohibited due to intellectual property and privacy concerns. Thus, there is a pressing need for encryption methods that obscure confidential aspects of the data, without affecting the outcomes of certain statistical analyses. A homomorphic encryption method for genotypes and phenotypes (HEGP) has been proposed for single-marker regression in genome-wide association studies (GWAS) using linear mixed models with Gaussian errors. This methodology permits frequentist likelihood-based parameter estimation and inference. In this paper, we extend HEGP to broader applications in genome-to-phenome analyses. We show that HEGP is suited to commonly used linear mixed models for genetic analyses of quantitative traits including genomic best linear unbiased prediction (GBLUP) and ridge-regression best linear unbiased prediction (RR-BLUP), as well as Bayesian variable selection methods (e.g. those in Bayesian Alphabet), for genetic parameter estimation, genomic prediction, and GWAS. By advancing the capabilities of HEGP, we offer researchers and industry professionals a secure and efficient approach for collaborative genomic analyses while preserving data confidentiality.

Genetic analysis of the blood transcriptome of young healthy pigs to improve disease resilience

BACKGROUND

Disease resilience is the ability of an animal to maintain productive performance under disease conditions and is an important selection target. In pig breeding programs, disease resilience must be evaluated on selection candidates without exposing them to disease. To identify potential genetic indicators for disease resilience that can be measured on selection candidates, we focused on the blood transcriptome of 1594 young healthy pigs with subsequent records on disease resilience. Transcriptome data were obtained by 3'mRNA sequencing and genotype data were from a 650 K genotyping array.

RESULTS

Heritabilities of the expression of 16,545 genes were estimated, of which 5665 genes showed significant estimates of heritability (p < 0.05), ranging from 0.05 to 0.90, with or without accounting for white blood cell composition. Genes with heritable expression levels were spread across chromosomes, but were enriched in the swine leukocyte antigen region (average estimate > 0.2). The correlation of heritability estimates with the corresponding estimates obtained for genes expressed in human blood was weak but a sizable number of genes with heritable expression levels overlapped. Genes with heritable expression levels were significantly enriched for biological processes such as cell activation, immune system process, stress response, and leukocyte activation, and were involved in various disease annotations such as RNA virus infection, including SARS-Cov2, as well as liver disease, and inflammation. To estimate genetic correlations with disease resilience, 3205 genotyped pigs, including the 1594 pigs with transcriptome data, were evaluated for disease resilience following their exposure to a natural polymicrobial disease challenge. Significant genetic correlations (p < 0.05) were observed with all resilience phenotypes, although few exceeded expected false discovery rates. Enrichment analysis of genes ranked by estimates of genetic correlations with resilience phenotypes revealed significance for biological processes such as regulation of cytokines, including interleukins and interferons, and chaperone mediated protein folding.

CONCLUSIONS

These results suggest that expression levels in the blood of young healthy pigs for genes in biological pathways related to immunity and endoplasmic reticulum stress have potential to be used as genetic indicator traits to select for disease resilience.

The WUR0000125 PRRS resilience SNP had no apparent effect on pigs' infectivity and susceptibility in a novel transmission trial

BACKGROUND

Porcine reproductive and respiratory syndrome (PRRS) remains one of the most important infectious diseases for the pig industry. A novel small-scale transmission experiment was designed to assess whether the WUR0000125 (WUR for Wageningen University and Research) PRRS resilience single nucleotide polymorphism (SNP) confers lower susceptibility and infectivity to pigs under natural porcine reproductive and respiratory syndrome virus (PRRSV-2) transmission.

METHODS

Commercial full- and half-sib piglets (n = 164) were assigned as either Inoculation, Shedder, or Contact pigs. Pigs were grouped according to their relatedness structure and WUR genotype, with R- and R+ referring to pigs with zero and one copy of the dominant WUR resilience allele, respectively. Barcoding of the PRRSV-2 strain (SD09-200) was applied to track pig genotype-specific transmission. Blood and nasal swab samples were collected and concentrations of PRRSV-2 were determined by quantitative (q)-PCR and cell culture and expressed in units of median tissue culture infectious dose (TCID₅₀). The Log₁₀TCID₅₀ at each sampling event, derived infection status, and area under the curve (AUC) were response variables in linear and generalized linear mixed models to infer WUR genotype differences in Contact pig susceptibility and Shedder pig infectivity.

RESULTS

All Shedder and Contact pigs, except one, became infected through natural transmission. There was no significant (p > 0.05) effect of Contact pig genotype on any virus measures that would indicate WUR genotype differences in susceptibility. Contact pigs tended to have higher serum AUC (p = 0.017) and log₁₀TCID₅₀ (p = 0.034) when infected by an R+ shedder, potentially due to more infectious R+ shedders at the early stages of the transmission trial. However, no significant Shedder genotype effect was found in serum (p = 0.274) or nasal secretion (p = 0.951) that would indicate genotype differences in infectivity.

CONCLUSIONS

The novel design demonstrated that it is possible to estimate genotype effects on Shedder pig infectivity and Contact pig susceptibility that are not confounded by family effects. The study, however, provided no supportive evidence that genetic selection on WUR genotype would affect PRRSV-2 transmission. The results of this study need to be independently validated in a larger trial using different PRRSV strains before dismissing the effects of the WUR marker or the previously detected GBP5 gene on PRRSV transmission.

Host response to successive challenges with lentogenic and velogenic Newcastle disease virus in local chickens of Ghana

Newcastle disease (ND) is a highly contagious viral disease that constantly threatens poultry production. The velogenic (highly virulent) form of ND inflicts the most damage and can lead to 100% mortality in unvaccinated village chicken flocks. This study sought to characterize responses of local chickens in Ghana after challenging them with lentogenic and velogenic Newcastle disease virus (NDV) strains. At 4 wk of age, chicks were challenged with lentogenic NDV. Traits measured were pre- and post-lentogenic infection growth rates (GR), viral load at 2 and 6 d post-lentogenic infection (DPI), viral clearance rate and antibody levels at 10 DPI. Subsequently, the chickens were naturally exposed to velogenic NDV (vNDV) after anti-NDV antibody titers had waned to levels ≤1:1,700. Body weights and blood samples were again collected for analysis. Finally, chickens were euthanized and lesion scores (LS) across tissues were recorded. Post-velogenic exposure GR; antibody levels at 21 and 34 days post-velogenic exposure (DPE); LS for trachea, proventriculus, intestines, and cecal tonsils; and average LS across tissues were measured. Variance components and heritabilities were estimated for all traits using univariate animal models. Mean pre- and post-lentogenic NDV infection GRs were 6.26 g/day and 7.93 g/day, respectively, but mean post-velogenic NDV exposure GR was −1.96 g/day. Mean lesion scores ranged from 0.52 (trachea) to 1.33 (intestine), with males having significantly higher (P < 0.05) lesion scores compared to females. Heritability estimates for the lentogenic NDV trial traits ranged from moderate (0.23) to high (0.55) whereas those for the vNDV natural exposure trial were very low (≤ 0.08). Therefore, in contrast to the vNDV exposure trial, differences in the traits measured in the lentogenic challenge were more affected by genetics and thus selection for these traits may be more feasible compared to those following vNDV exposure. Our results can form the basis for identifying local chickens with improved resilience in the face of NDV infection for selective breeding to improve productivity.

Effect of the host genotype at a Porcine Reproductive and Respiratory Syndrome (PRRS) resistance marker on evolution of the modified-live PRRS vaccine virus in pigs

Porcine Reproductive and Respiratory Syndrome, Porcine Reproductive and Respiratory Syndrome Virus, PRRS, Whole genome sequencing, quasispecies, WUR allele.

Complex immunogenomic control of serum thyroid hormone levels in fetuses and piglets challenged with Porcine Reproductive and Respiratory Syndrome Virus

Porcine Reproductive and Respiratory Syndrome Virus (PRRSV) causes respiratory disease in piglets and reproductive disease in sows. However, the interaction between host immunity and hormone homeostasis is not completely understood but piglet and fetal serum thyroid hormone (i.e., T3 and T4) levels decrease rapidly in response to PRRSV infection. Our objective was to estimate genetic parameters, identify QTL, and uncover nearby candidate genes for T3 and/or T4 levels of piglets and fetuses challenged with PRRSV2. Serum from 5-week old pigs (N=1792) at 11 days post infection (dpi) with PRRSV2 were assayed for T3 levels (piglet_T3). While serum from fetuses (N=1267) at 12 or 21 dpi from sows (N=150) challenged with PRRSV2 in late gestation were assayed for T3 (fetal_T3) and T4 (fetal_T4) levels. Animals were genotyped for 60K or 650K SNPs, heritabilities were estimated and genome wide association studies were performed for each trait separately. We found all three traits to be moderately heritable (18–28%) and the significant QTL (N=37) across several chromosomes collectively explained 24–34% of the genetic variation. A shared QTL identified for all traits on chromosome 6 contained the IL12RB2 and IFI44 genes. The largest QTL identified for both piglet_T3 and fetal_T3 on chromosome 5 contained IL17RA. A large QTL identified for piglet_T3 on chromosome 7 contained 9 MHC genes (e.g., SLA-DMA, SLA-DOA, SLA-DQA, SLA-DRA, and MHC-IA7). In conclusion, our data support the concept that there are complex interactions between thyroid hormone levels and multiple immunogenetic pathways during PRRSV infection, and that there is potential to select for pleiotropic QTL to simultaneously improve host immunity for reproductive and respiratory PRRS.

Genetic parameters and purebred-crossbred genetic correlations for growth, meat quality, and carcass traits in pigs

Growth, meat quality, and carcass traits are of economic importance in swine breeding. Understanding their genetic basis in purebred (PB) and commercial crossbred (CB) pigs is necessary for a successful breeding program because, although the breeding goal is to improve CB performance, phenotype collection and selection are usually carried out in PB populations housed in biosecure nucleus herds. Thus, the selection is indirect, and the accuracy of selection depends on the genetic correlation between PB and CB performance (rpc). The objectives of this study were to 1) estimate genetic parameters for growth, meat quality, and carcass traits in a PB sire line and related commercial CB pigs and 2) estimate the corresponding genetic correlations between purebred and crossbred performance (rpc). Both objectives were investigated by using pedigree information only (PBLUP) and by combining pedigree and genomic information in a single-step genomic BLUP (ssGBLUP) procedure. Growth rate showed moderate estimates of heritability for both PB and CB based on PBLUP, while estimates were higher in CB based on ssGBLUP. Heritability estimates for meat quality traits were diverse and slightly different based on PB and CB data with both methods. Carcass traits had higher heritability estimates based on PB compared with CB data based on PBLUP and slightly higher estimates for CB data based on ssGBLUP. A wide range of estimates of genetic correlations were obtained among traits within the PB and CB data. In the PB population, estimates of heritabilities and genetic correlations were similar based on PBLUP and ssGBLUP for all traits, while based on the CB data, ssGBLUP resulted in different estimates of genetic parameters with lower SEs. With some exceptions, estimates of rpc were moderate to high. The SE on the rpc estimates was generally large when based on PBLUP due to limited sample size, especially for CBs. In contrast, estimates of rpc based on ssGBLUP were not only more precise but also more consistent among pairs of traits, considering their genetic correlations within the PB and CB data. The wide range of estimates of rpc (less than 0.70 for 7 out of 13 traits) indicates that the use of CB phenotypes recorded on commercial farms, along with genomic information, for selection in the PB population has potential to increase the genetic progress of CB performance.

93 Genomic basis of antibody response to porcine reproductive and respiratory syndrome virus vaccination

Antibody (Ab) response to natural infection with porcine reproductive and respiratory syndrome (PRRS) virus has been shown to be highly heritable (~0.40), to be genetically correlated with farrowing performance in PRRSV-infected sows, be controlled by two major QTL on chromosome 7, among others, and have moderate genomic prediction accuracy. However, waiting for PRRS outbreaks to occur to collect data limits the use of Ab response to select for increased PRRS resilience. Thus, we investigated the genomic basis of Ab response to PRRS vaccination with a modified live PRRSV vaccine as a strategy to generate data for this purpose. Nine hundred and six commercial F1 replacement gilts (189±16 days old) were vaccinated with a commercial PRRS modified live virus vaccine. Blood samples were collected 52 days after vaccination to measure Ab response, as sample-to-positive (S/P) ratio using a commercial ELISA, and for SNP genotyping (~50K). BayesC0 was used to estimate heritability for S/P ratio using in a model with contemporary group (CG) as fixed effect and SNP effects as random. Genome-wide association study and genomic prediction for S/P ratio were performed with BayesB (Pi=0.99). For genomic prediction, a three-fold cross-validation was used, in which each CG (n=3) was used as validation dataset. Accuracy of genomic prediction was defined as the correlation between genomic estimated breeding values and phenotypes adjusted for estimates of fixed effects, weighed by the number of individuals in the validation dataset. Heritability of S/P was moderate (0.35±0.04). A QTL was identified on chromosome 7 (25 Mb) explaining ~28% of the genetic variance. Accuracy of genomic prediction was fairly high (0.60±0.15). This is the first study describing the genomic basis of Ab response to PRRS vaccination with modified-live virus vaccine. Additional work is needed to evaluate the genetic correlation of Ab response to vaccination with resilience traits in pigs.

154 A genetic marker for PRRS resistance has no adverse effect on economically important traits in pigs

Porcine Respiratory and Reproductive Syndrome (PRRS) is a very costly viral disease of pigs. A genetic marker for host response to PRRS on chromosome 4 (rs80800372, (WUR)) could be used to select for resilience to PPRS. However, before including WUR in selection, it is necessary to know effects of WUR on economically important traits without PRRS. A dataset with both genotypes and phenotypes on purebred Duroc (D), Landrace (L), Yorkshire (Y) and crossbred (LY or DLY) pigs was used. A total of 20 traits were tested. Sow reproduction and litter traits were available on &gt; 13,000 D, L and Y and 1,549 LY sows. Growth and ultrasound traits were recorded on &gt;35,000 D, L and Y and 2,622 DLY pigs. Daily feed intake (70-120kg) was from 4,133 Durocs. Carcass and meat quality traits were from 2,184 Durocs, 1,160 Yorkshires and 2,184 DLY pigs. All animals were genotyped using a custom SNP chip ( &gt; 55K) including WUR. Analyses were done within each breed using an animal model. WUR was fitted as a fixed effect by classifying pigs with one or two favourable alleles (B) into one group and others into a second group. Contemporary group, parity, litter, lactation length and net fostering effects were considered depending on the traits. Relationship matrices were constructed using pedigree in purebreds and genotypes in crossbreds. WUR had no significant ( P &gt;0.05 ) effect on any trait, except for number of pigs alive at 24hrs in Y and ultrasound loin depth in D and Y ( P&lt; 0.05 ) but the favorable WUR allele also had favorable effects on these two traits. These results were similar to the report by Dunkelberger et al. (J. Anim. Sci. 2017, 95: 2838). In conclusion, WUR had no adverse effects on any trait and can be used to select pigs with increased resilience to PRRS.

PSV-11 Accuracies of genomic prediction for reproductive traits in PRRSV-infected sows

Recent studies on host-genetics of response to porcine reproductive and respiratory syndrome virus (PRRSV) infection have shown that genetic markers can be used to predict performance in growing pigs. On the other hand, no work has been done on the genomic prediction of reproductive performance in PRRSV-infected sows. Thus, the objective of this work was to assess the accuracies of genomic prediction for reproductive traits in PRRSV-infected sows. A total of 475 Duroc and 405 Landrace sows with ~30K SNP genotypes had farrowing performance data during the outbreak on: number of piglets born alive (NBA), stillborn piglets (NSB), mummified piglets (NM), piglets born dead (NBD; NSB+NM), total number of piglets born (TNB; NBA+NBD), and piglets weaned (NW). Genomic prediction was performed by Bayes-B with a model fitting parity as fixed-effect and SNP effects as random. Analyses were performed for two scenarios: within-breed and between-breed. For the within-breed prediction, 4-fold cross-validation was used. For the between-breed prediction, all data for each breed were used. Accuracy of genomic prediction (AGP) was calculated as the correlation between genomic estimated breeding values and pre-adjusted phenotype, divided by square-root of heritability. For the within-breed prediction, AGP (± standard deviation) for NBA, NBD, NM, NSB, NW, and TNB were 0.24±0.05, 0.07±0.10, -0.27±0.06, -0.24±0.09, -0.05±0.05, 0.19±0.03, respectively, for Duroc, and 0.37±0.02, 0.38±0.11, 0.30±0.07, 0.60±0.10, 0.09±0.11, 0.15±0.10, respectively, for Landrace. For the between-breed prediction, AGP for NBA, NBD, NM, NSB, NW, and TB were 0.28, -0.17, -0.01, 0.37, 0.27, and 0.29, respectively, when training on Landrace and validating on Duroc, and 0.28, -0.14, 0.01, 0.15, 0.23, and -0.01, respectively, when training on Duroc and validating on Landrace. Results show that the accuracies of genomic prediction for reproductive traits during a PRRSV outbreak are low to moderate.

PSV-15 Genomic basis of reproductive performance in PRRSV-infected sows

Previous studies have shown that reproductive traits in sows during porcine reproductive and respiratory syndrome (PRRS) outbreaks are lowly heritable. However, little is known about the genomic basis of reproductive performance in PRRS virus (PRRSV)-infected sows. Thus, the objective of this work was to evaluate the genomic basis of reproductive performance in PRRSV-infected sows. Farrowing data from 475 Duroc and 405 Landrace sows that experienced a PRRS outbreak were used. Heritability and genome-wide association studies (GWAS) based on ~30K SNP genotypes were performed for each breed separately, for number of piglets born alive (NBA), stillborn piglets (NSB), mummified piglets (NM), piglets born dead (NBD; NSB+NM), total number of piglets born (TNB; NBA+NBD), and piglets weaned (NW). Heritabilities were estimated with ASReml using a model with the fixed-effect of parity and an animal random-effect, using a genomic-relationship matrix. GWAS was performed using Bayes-B in GenSel with a model fitting parity as fixed-effect and SNP effects as random. Heritability estimates (± standard error) for NBA, NSB, NM, NBD, TNB, and NW were 0.07±0.07, 0.02±0.11, 0.04±0.10, 0.08±0.08, 0.08±0.07, 0.06±0.04, respectively, for Duroc, and 0.09±0.07, 0.10±0.07, 0.10±0.08, 0.11±0.07, 0.07±0.08, 0.07±0.08, respectively, for Landrace. For Duroc, GWAS identified a quantitative trait locus (QTL) on chromosome (chr) 3 [114-120 megabases (Mb)] explaining 3.9% of the genetic variance (GV) for NBA, on chr 13 (189-193 Mb) explaining 2.7% GV for NBD, and on chr 5 (36-41 Mb) explaining 4.3% GV for TNB. For Landrace, GWAS identified a QTL on chr 13 (154-162 Mb) explaining 12.8% GV for NBA, on chr 14 (59-69 Mb) explaining 2.5% GV for NBD, and on chr 10 (7-9 Mb) explaining 4.1% GV for NW. These results indicate that reproductive traits are lowly heritable during a PRRS outbreak with few QTL identified in Duroc and Landrace sows.

Phenotypic variability and population structure analysis of Tanzanian free-range local chickens

BACKGROUND

Free-range local chickens (FRLC) farming is an important activity in Tanzania, however, they have not been well-characterized. This study aimed to phenotypically characterize three Tanzanian FRLCs and to determine their population structure. A total of 389 mature breeder chickens (324 females and 65 males) from three popular Tanzanian FRLC ecotypes (Kuchi, Morogoro-medium and Ching'wekwe) were used for the phenotypic characterization. Progenies of these chickens were utilized to assess population structure. The ecotypes were collected from four geographical zones across Tanzania: Lake, Central, Northern and Coastal zones. Body weights and linear measurements were obtained from the mature breeders, including body, neck, shanks, wingspan, chest girth, and shank girth. Descriptive statistics were utilized to characterize the chickens. Correlations between the linear measurements and differences among the means of measured linear traits between ecotypes and between sexes were assessed. A total of 1399 progeny chicks were genotyped using a chicken 600 K high density single nucleotide polymorphism (SNP) panel for determination of population structure.

RESULTS

The means for most traits were significantly higher in Kuchi relative to Ching'wekwe and Morogoro-medium. However, shank length and shank girth were similar between Kuchi and Morogoro-medium females. All traits were correlated with the exception of shank girth in Morogoro-medium. Admixture analyses revealed that Morogoro-medium and Ching'wekwe clustered together as one population, separate from Kuchi.

CONCLUSIONS

Phenotypic traits could be used to characterize FRLCs, however, there were variations in traits among individuals within ecotypes; therefore, complementary genomic methods should be considered to improve the characterization for selective breeding.

Responsiveness of swine divergently selected for feed efficiency to exogenous adrenocorticotropic hormone and glucose challenges

Increasing the feed efficiency of lean tissue gains is an important goal for improving sustainable pork production and profitability for swine producers. To study feed efficiency, genetic selection based on residual feed intake (RFI) was used to create two divergent lines. Low-RFI pigs consume less feed for equal weight gain compared with their less-efficient, high-RFI counterparts. As cortisol and insulin are important energy control and growth regulators, our objective was to evaluate the role of the adrenocorticotropic hormone (ACTH)-cortisol and the glucose-insulin axes in pigs divergently selected for RFI. Adrenocorticotropic hormone (0.2 IU/kg BW)-stimulated cortisol and non-esterified fatty acids (NEFA) concentrations and intravenous glucose tolerance test (IVGTT; 0.25 g/kg BW)-stimulated glucose, insulin, and NEFA concentrations were assessed in six low-RFI and six high-RFI gilts (68 ± 5.2 kg). Before the ACTH challenge, low-RFI gilts tended to have less baseline plasma cortisol (P = 0.08) but no difference in NEFA concentrations (P = 0.63) compared with high-RFI gilts. After the ACTH challenge, low-RFI gilts had less cortisol (P = 0.04) and NEFA concentrations (P = 0.05) compared with high-RFI gilts. Glucose, insulin, and NEFA concentrations did not differ between genetic lines before the IVGTT. After glucose infusion, low-RFI gilts had greater insulin concentrations (P = 0.003) but did not differ in glucose or NEFA concentrations compared with high-RFI gilts. These results indicate that genetic selection for reduced RFI (improved feed efficiency) resulted in less stress responsiveness and an increase in insulin after glucose infusion. These data have implications for identifying and selecting more feed efficient pigs and for understanding the physiological mechanisms underlying feed efficiency.

Creating effective biocontainment facilities and maintenance protocols for raising specific pathogen-free, severe combined immunodeficient (SCID) pigs

Severe combined immunodeficiency (SCID) is defined by the lack of an adaptive immune system. Mutations causing SCID are found naturally in humans, mice, horses, dogs, and recently in pigs, with the serendipitous discovery of the Iowa State University SCID pigs. As research models, SCID animals are naturally tolerant of xenotransplantation and offer valuable insight into research areas such as regenerative medicine, cancer therapy, as well as immune cell signaling mechanisms. Large-animal biomedical models, particularly pigs, are increasingly essential to advance the efficacy and safety of novel regenerative therapies on human disease. Thus, there is a need to create practical approaches to maintain hygienic severe immunocompromised porcine models for exploratory medical research. Such research often requires stable genetic lines for replication and survival of healthy SCID animals for months post-treatment. A further hurdle in the development of the ISU SCID pig as a biomedical model involved the establishment of facilities and protocols necessary to obtain clean SPF piglets from the conventional pig farm on which they were discovered. A colony of homozygous SCID boars and SPF carrier sows has been created and maintained through selective breeding, bone marrow transplants, innovative husbandry techniques, and the development of biocontainment facilities.

Genomic evaluation of feed efficiency component traits in Duroc pigs using 80K, 650K and whole-genome sequence variants

BACKGROUND

Increasing marker density was proposed to have potential to improve the accuracy of genomic prediction for quantitative traits; whole-sequence data is expected to give the best accuracy of prediction, since all causal mutations that underlie a trait are expected to be included. However, in cattle and chicken, this assumption is not supported by empirical studies. Our objective was to compare the accuracy of genomic prediction of feed efficiency component traits in Duroc pigs using single nucleotide polymorphism (SNP) panels of 80K, imputed 650K, and whole-genome sequence variants using GBLUP, BayesB and BayesRC methods, with the ultimate purpose to determine the optimal method to increase genetic gain for feed efficiency in pigs.

RESULTS

Phenotypes of average daily feed intake (ADFI), average daily gain (ADG), ultrasound backfat depth (FAT), and loin muscle depth (LMD) were available for 1363 Duroc boars from a commercial breeding program. Genotype imputation accuracies reached 92.1% from 80K to 650K and 85.6% from 650K to whole-genome sequence variants. Average accuracies across methods and marker densities of genomic prediction of ADFI, FAT, LMD and ADG were 0.40, 0.65, 0.30 and 0.15, respectively. For ADFI and FAT, BayesB outperformed GBLUP, but increasing marker density had little advantage for genomic prediction. For ADG and LMD, GBLUP outperformed BayesB, while BayesRC based on whole-genome sequence data gave the best accuracies and reached up to 0.35 for LMD and 0.25 for ADG.

CONCLUSIONS

Use of genomic information was beneficial for prediction of ADFI and FAT but not for that of ADG and LMD compared to pedigree-based estimates. BayesB based on 80K SNPs gave the best genomic prediction accuracy for ADFI and FAT, while BayesRC based on whole-genome sequence data performed best for ADG and LMD. We suggest that these differences between traits in the effect of marker density and method on accuracy of genomic prediction are mainly due to the underlying genetic architecture of the traits.

A nested mixture model for genomic prediction using whole-genome SNP genotypes

Genomic prediction exploits single nucleotide polymorphisms (SNPs) across the whole genome for predicting genetic merit of selection candidates. In most models for genomic prediction, e.g. BayesA, B, C, R and GBLUP, independence of SNP effects is assumed. However, SNP effects are expected to be locally dependent given the presence of a nearby QTL because SNPs surrounding the QTL do not segregate independently. A consequence of ignoring this dependence is that SNPs with small effects may be overly shrunk, e.g. effects from markers with high minor allele frequencies (MAF) that flank QTL with low MAF. A nested mixture model (BayesN) is developed to account for the dependence of effects of SNPs that are closely linked, where the effects of SNPs in every non-overlapping genomic window a priori follow a point mass at zero for all SNPs or a mixture of some SNPs with nonzero effects and others with zero effects. It can be regarded as a parsimonious alternative to the existing antedependence model, antiBayesB, which allow a nonstationary dependence of SNP effects. Illumina 777K BovineHD genotypes from 948 Angus cattle were used to simulate 5,000 offspring, with 4,000 used for training and 1,000 for validation. Scenarios with 300 common (MAF > 0.05) or rare (MAF < 0.05) QTL randomly selected from segregating SNPs were replicated 8 times. SNPs corresponding to QTL were masked from a 600k panel comprising SNPs with MAF > 0.05 or a 50k evenly spaced subset of these. Compared with BayesB and a modified antiBayesB, BayesN improved the accuracy of prediction up to 2.0% with 50k SNPs and up to 7.0% with 600k SNPs, most improvements occurring in the rare QTL scenario. Computing time was reduced up to 60% with 50k SNPs and up to 75% with 600k SNPs. BayesN is an accurate and computationally efficient method for genomic prediction with whole-genome SNPs, especially for traits with rare QTL.

Comparison of alternative approaches to single-trait genomic prediction using genotyped and non-genotyped Hanwoo beef cattle

BACKGROUND

Genomic predictions from BayesA and BayesB use training data that include animals with both phenotypes and genotypes. Single-step methodologies allow additional information from non-genotyped relatives to be included in the analysis. The single-step genomic best linear unbiased prediction (SSGBLUP) method uses a relationship matrix computed from marker and pedigree information, in which missing genotypes are imputed implicitly. Single-step Bayesian regression (SSBR) extends SSGBLUP to BayesB-like models using explicitly imputed genotypes for non-genotyped individuals.

METHODS

Carcass records included 988 genotyped Hanwoo steers with 35,882 SNPs and 1438 non-genotyped steers that were measured for back-fat thickness (BFT), carcass weight (CWT), eye-muscle area, and marbling score (MAR). Single-trait pedigree-based BLUP, Bayesian methods using only genotyped individuals, SSGBLUP and SSBR methods were compared using cross-validation.

RESULTS

Methods using genomic information always outperformed pedigree-based BLUP when the same phenotypic data were modeled from either genotyped individuals only or both genotyped and non-genotyped individuals. For BFT and MAR, accuracies were higher with single-step methods than with BayesB, BayesC and BayesCπ. Gains in accuracy with the single-step methods ranged from +0.06 to +0.09 for BFT and from +0.05 to +0.07 for MAR. For CWT, SSBR always outperformed the corresponding Bayesian methods that used only genotyped individuals. However, although SSGBLUP incorporated information from non-genotyped individuals, prediction accuracies were lower with SSGBLUP than with BayesC (π = 0.9999) and BayesB (π = 0.98) for CWT because, for this particular trait, there was a benefit from the mixture priors of the effects of the single nucleotide polymorphisms.

CONCLUSIONS

Single-step methods are the preferred approaches for prediction combining genotyped and non-genotyped animals. Alternative priors allow SSBR to outperform SSGBLUP in some cases.

Contributions of linkage disequilibrium and co-segregation information to the accuracy of genomic prediction

Background

Traditional genomic prediction models using multiple regression on single nucleotide polymorphisms (SNPs) genotypes exploit associations between genotypes of quantitative trait loci (QTL) and SNPs, which can be created by historical linkage disequilibrium (LD), recent co-segregation (CS) and pedigree relationships. Results from field data analyses show that prediction accuracy is usually much higher for individuals that are close relatives of the training population than for distantly related individuals. A possible reason is that historical LD between QTL and SNPs is weak and, for close relatives, prediction accuracy of SNP models is mainly contributed by pedigree relationships and CS. Information from pedigree relationships decreases fast over generations and only contributes to within-family prediction. Information from CS is affected by family structures and effective population size, and can have a substantial contribution to prediction accuracy when modeled explicitly.

Results

In this study, a method to explicitly model CS was developed by following the transmission of putative QTL alleles using allele origins at SNPs. Bayesian hierarchical models that combine information from LD and CS (LD-CS model) were developed for genomic prediction in pedigree populations. Contributions of LD and CS information to prediction accuracy across families and generations without retraining were investigated in simulated half-sib datasets and deep pedigrees with different recent effective population sizes, respectively. Results from half-sib datasets showed that when historical LD between QTL and SNPs is low, accuracy of the LD model decreased when the training data size is increased by adding independent sire families, but accuracies from the CS and LD-CS models increased and plateaued rapidly. Results from deep pedigree datasets show that the LD model had high accuracy across generations only when historical LD between QTL and SNPs was high. Modeling CS explicitly resulted in higher accuracy than the LD model across generations when the mating design generated many close relatives.

Conclusions

Our results suggest that modeling CS explicitly improves accuracy of genomic prediction when historical LD between QTL and SNPs is low. Modeling both LD and CS explicitly is expected to improve accuracy when recent effective population size is small, or when the training data include many independent families.

Electronic supplementary material

The online version of this article (doi:10.1186/s12711-016-0255-4) contains supplementary material, which is available to authorized users.

Genetics Selection Evolution reviewer acknowledgement 2014

The Genetics Selection Evolution editorial team would sincerely like to thank all of our reviewers who contributed to peer review for the journal in 2014.

Adaptation of novel H7N9 influenza A virus to human receptors

The emergence of the novel H7N9 influenza A virus (IAV) has caused global concerns about the ability of this virus to spread between humans. Analysis of the receptor-binding properties of this virus using a recombinant protein approach in combination with fetuin-binding, glycan array and human tissue-binding assays demonstrates increased binding of H7 to both α2-6 and α2-8 sialosides as well as reduced binding to α2-3-linked SIAs compared to a closely related avian H7N9 virus from 2008. These differences could be attributed to substitutions Q226L and G186V. Analysis of the enzymatic activity of the neuraminidase N9 protein indicated a reduced sialidase activity, consistent with the reduced binding of H7 to α2-3 sialosides. However, the novel H7N9 virus still preferred binding to α2-3- over α2-6-linked SIAs and was not able to efficiently bind to epithelial cells of human trachea in contrast to seasonal IAV, consistent with its limited human-to-human transmission.

Characterization of a naturally occurring severe combined immunodeficiency in pigs (P6087)

Severe combined immunodeficiency (SCID) refers to a group of inherited disorders in which two components of the immune system are defective. We have identified a naturally occurring, SCID in a line of pigs that were bred for increased feed efficiency. Affected piglets remain healthy while on the sow, but succumb to infections soon after weaning. SCID piglets have underdeveloped lymph nodes and spleen with a paucity of lymphocytes, they do not make antibodies to Porcine Reproductive and Respiratory Syndrome (PRRS) virus, and they accept human tumors. Further analyses showed that the defect has an autosomal recessive inheritance pattern. Blood was collected within 2 days after birth to titer serum immunoglobulins, perform lymphocyte counts and identify lymphocyte populations by flow cytometry. Affected piglets had no detectable serum immunoglobulins, and very low numbers of circulating lymphocytes. By flow cytometry, there were Natural Killer cells, but few to no T or B lymphocytes. At necropsy, thymus and small intestine were collected from each affected piglet and from control littermates. Histologic sections, stained with Hematoxylin and Eosin, were evaluated for cellularity. Affected piglet thymic remnants were small, containing Hassall’s corpuscles, but relatively few thymocyte-like cells. There were no detectable Peyer’s Patches in small intestine. These observations are consistent with T-/B- NK+ SCID.

Control of porcine reproductive and respiratory syndrome (PRRS) through genetic improvements in disease resistance and tolerance

Infections caused by porcine reproductive and respiratory syndrome virus (PRRSV) have a severe economic impact on pig production in North America, Europe, and Asia. The emergence and eventual predominance of PRRS in the 1990s are the likely result of changes in the pork industry initiated in the late 1970s, which allowed the virus to occupy a unique niche within a modern commercial production system. PRRSV infection is responsible for severe clinical disease, but can maintain a life-long subclinical infection, as well as participate in several polymicrobial syndromes. Current vaccines lessen clinical signs, but are of limited use for disease control and elimination. The relatively poor protective immunity following vaccination is a function of the virus's capacity to generate a large degree of genetic diversity, combined with several strategies to evade innate and adaptive immune responses. In 2007, the PRRS Host Genetics consortium (PHGC) was established to explore the role of host genetics as an avenue for PRRS control. The PHGC model for PRRS incorporates the experimental infection of large numbers of growing pigs and has created the opportunity to study experimental PRRSV infection at the population level. The results show that pigs can be placed into distinct phenotypic groups, including pigs that show resistance (i.e., low virus load) or pigs that exhibit “tolerance” to infection. Tolerance was illustrated by pigs that gain weight normally in the face of a relatively high virus load. Genome-wide association analysis has identified a region on chromosome 4 (SSC4) correlated with resistance; i.e., lower cumulative virus load within the first 42 days of infection. The genomic region is near a family of genes involved in innate immunity. The region is also associated with higher weight gain in challenged pigs, suggesting that pigs with the resistance alleles don't seem to simultaneously experience reduction in growth, i.e., that resistance and tolerance are not antagonistically related. These results create the opportunity to develop breeding programs that will produce pigs with increased resistance to PRRS and simultaneously high growth rate. The identification of genomic markers involved in actual tolerance will likely prove more difficult, primarily because tolerance is difficult to quantify and because tolerance mechanism are still poorly understood. Another avenue of study includes the identification of genomic markers related to improved response following vaccination.

ANEXdb: an integrated animal ANnotation and microarray EXpression database

To determine annotations of the sequence elements on microarrays used for transcriptional profiling experiments in livestock species, currently researchers must either use the sparse direct annotations available for these species or create their own annotations. ANEXdb ( http://www.anexdb.org ) is an open-source web application that supports integrated access of two databases that house microarray expression (ExpressDB) and EST annotation (AnnotDB) data. The expression database currently supports storage and querying of Affymetrix-based expression data as well as retrieval of experiments in a form ready for NCBI-GEO submission; these services are available online. AnnotDB currently houses a novel assembly of approximately 1.6 million unique porcine-expressed sequence reads called the Iowa Porcine Assembly (IPA), which consists of 140,087 consensus sequences, the Iowa Tentative Consensus (ITC) sequences, and 103,888 singletons. The IPA has been annotated via transfer of information from homologs identified through sequence alignment to NCBI RefSeq. These annotated sequences have been mapped to the Affymetrix porcine array elements, providing annotation for 22,569 of the 23,937 (94%) porcine-specific probe sets, of which 19,253 (80%) are linked to an NCBI RefSeq entry. The ITC has also been mined for sequence variation, providing evidence for up to 202,383 SNPs, 62,048 deletions, and 958 insertions in porcine-expressed sequence. These results create a single location to obtain porcine annotation of and sequence variation in differently expressed genes in expression experiments, thus permitting possible identification of causal variants in such genes of interest. The ANEXdb application is open source and available from SourceForge.net.

Over-expression of parvalbumin in transgenic mice rescues motoneurons from injury-induced cell death

Following nerve injury in neonatal rats, a large proportion of motoneurons die, possibly as a consequence of an increase in vulnerability to the excitotoxic effects of glutamate. Calcium-dependent glutamate excitotoxicity is thought to play a significant role not only in injury-induced motoneuron death, but also in motoneuron degeneration in diseases such as amyotrophic lateral sclerosis (ALS). Motoneurons are particularly vulnerable to calcium influx following glutamate receptor activation, as they lack a number of calcium binding proteins, such as calbindin-D(28k) and parvalbumin. Therefore, it is possible that increasing the ability of motoneurons to buffer intracellular calcium may protect them from cell death and prevent the decline in motor function that usually occurs as a consequence of motoneuron loss. In this study we have tested this possibility by examining the effect of neonatal axotomy on motoneuron survival and muscle force production in normal and transgenic mice that over-express parvalbumin in their motoneurons.The sciatic nerve was crushed in one hindlimb of new-born transgenic and wildtype mice. The effect on motoneuron survival was assessed 8 weeks later by retrograde labelling of motoneurons innervating the tibialis anterior muscle. Following nerve injury in wildtype mice, only 20.2% (+/-2.2, S.E.M.; n=4) of injured motoneurons survive long term compared with 47.2% (+/-4.4, S.E.M.; n=4) in parvalbumin over-expressing mice. Surprisingly, this dramatic increase in motoneuron survival was not reflected in a significant improvement in muscle function, since 8 weeks after injury there was no improvement in either maximal twitch and tetanic force, or muscle weights.Thus, inducing spinal motoneurons to express parvalbumin protects a large proportion of motoneurons from injury-induced cell death, but this is not sufficient to restore muscle function.

Treatment of the neuromuscular junction with 4-aminopyridine results in improved reinnervation following nerve injury in neonatal rats

During early postnatal development, nerve injury results in the death of a large proportion of motoneurones and poor recovery of muscle function. Our previous results have shown that premature enhancement of transmitter release from nerve terminals prevents the death of motoneurones following neonatal nerve injury. Whether this increase in motoneurone survival is reflected in an improvement in the reinnervation of muscle was studied here. The muscles in one hindlimb of newborn rats were treated with 4-aminopyridine. Three days later, the sciatic nerve was crushed in the treated leg. When the animals were seven, 14 and 21days of age, the soleus and extensor digitorum longus muscles were removed and processed for GAP-43 (a 43-kDa growth-associated protein) and synaptophysin immunocytochemistry. Both GAP-43 and synaptophysin were expressed in normal soleus and extensor digitorum longus muscles at seven days. Synaptophysin was still expressed at 14 days, but GAP-43 expression had declined. Following nerve injury at three days of age, there was no GAP-43 or synaptophysin immunoreactivity in nerve terminals at seven days. By 21 days, there were 17.3+/-2.1 GAP-43-positive terminals per section in the soleus and 17.7+/-1.4 in the extensor digitorum longus, with mean terminal areas of 47.5+/-3.3 and 49.8+/-2.6 microm(2), respectively. In animals in which nerve crush was preceded by 4-aminopyridine treatment, at 21 days there were 32.9+/-2.6 GAP-43-immunoreactive terminals in the soleus and 44.9+/-2.3 in the extensor digitorum longus, with a mean area of 122.7+/-6.6 microm(2) in the soleus and 136.2+/-9.7 microm(2) in the extensor digitorum longus. These results indicate that in muscles pretreated with 4-aminopyridine, prior to nerve crush at three days, there are significantly more terminals, which occupy a larger area than in untreated muscles. Thus, increasing transmitter release prior to nerve injury significantly improved the ability of axons to reinnervate muscle.

Changes in expression of NMDA receptor subunits in the rat lumbar spinal cord following neonatal nerve injury

The vulnerability of motoneurones to glutamate has been implicated in neurological disorders such as amyotrophic lateral sclerosis but it is not known whether specific receptor subtypes mediate this effect. In order to investigate this further, the expression of N-methyl-D-aspartate (NMDA) receptor subunits was studied during the first three post-natal weeks when motoneurones are differentially vulnerable to injury following neonatal nerve crush compared to the adult. Unilateral nerve crush was carried out at day 2 after birth (P2) which causes a decrease of 66% in motoneurone number by 14 days (P14). To study receptor expression in identified motoneurones, serial section analysis was carried out on retrogradely labelled common peroneal (CP) motoneurones by combined immunocytochemistry and in situ hybridization (ISH). mRNA levels were also quantified in homogenates from lumbar spinal cords in which the side ipsilateral to the crush was separated from the contralateral side. The NR1 subunit of the NMDA receptor was widely distributed in the spinal cord being expressed most strongly in motoneurone somata particularly during the neonatal period (P3-P7). The NR2 subunits were also expressed at higher levels in the somata and dendrites of neonatal motoneurones compared to older animals. NR2B mRNA was expressed at low to moderate levels throughout the studied period whereas NR2A mRNA levels were low until P21. Following unilateral nerve crush, an initial decrease in NR1 mRNA occurred at one day after nerve crush (P3) in labelled CP motoneurones ipsilateral to the crush which was followed by a significant increase in NR1 subunit expression at 5 days post-injury. This increase was bilateral although reaching greater significance ipsilateral to the crush compared with sham-operated animals. A significant increase in NR1 and NR2B mRNA post injury was also detected in spinal cord homogenates. In addition, the changes in levels of NR1 and NR2B mRNA were reflected by comparable bilateral changes at P7 in receptor protein determined by quantitative immunocytochemical analysis of NR1 and NR2 subunit expression in identified CP motoneurones indicating a co-ordinated regulation of receptor subunits in response to injury.

Persistence of somatic and dendritic growth associated processes and induction of dendritic sprouting in motoneurones after neonatal axotomy in the rat

The effect of neonatal axotomy on the maturation of motoneurone somadendritic morphology was studied in identified motoneurones innervating the ankle dorsiflexor muscles tibialis anterior (TA) and extensor digitorum longus (EDL) of the rat by intracellular injection of Lucifer Yellow and confocal microscopy. At birth, the entire somatodendritic surface is covered with fine filopodial growth-associated processes. These are eliminated from the soma and proximal dendrites during the first postnatal week as part of a somatofugal process of dendritic maturation. Following neonatal axotomy, the postnatal elimination of growth associated processes was halted and new, axonal-like processes were seen to sprout from the soma and proximal dendrites in some of the axotomized motoneurones. These results indicate that synaptic interaction with the target muscle during the early postnatal period is essential for the maturation of the somatodendritic receptive surface of the motoneurone.

Early postnatal changes in the somatodendritic morphology of ankle flexor motoneurons in the rat

The development of locomotor function in the rat spans the first 3 postnatal weeks. We have studied morphological features of the soma and dendrites of motoneurons innervating the physiological flexor muscles of the ankle, tibialis anterior and extensor digitorum longus, by intracellular injection in vitro between the first and ninth postnatal days. We obtained serial optical sections of 96 adequately filled motoneurons in whole-mounted hemisected spinal cords by confocal microscopy, projected them onto a single plane and analysed them morphometrically. On the day after birth, the somatodendritic surfaces of most such motoneurons were covered in growth-associated spiny, thorny or hair-like appendages. These had disappeared from the soma by the fourth postnatal day and from most proximal dendrites by day 7, but were still common distally on day 9. During this period there was little or no net growth of either the soma (which was still much smaller than in the adult) or the dendritic tree. A dorsal dendritic bias was present and 'sprays' of long, loosely bundled dorsal dendrites were often seen. The mean number of primary dendrites remained constant at about eight, and their combined diameter was already significantly correlated with mean soma diameter, as in the adult cat. Thus, the critical neonatal period during which these ankle flexor motoneurons are known to change their electrophysiological properties and to be particularly sensitive to interference with neuromuscular interaction is characterized by major changes in the neuronal surface, presumably linked to synaptogenesis.

Enhancing the laser scanning confocal microscopic visualization of Lucifer yellow filled cells in whole-mounted tissue

The laser scanning confocal microscope (LSCM) is an extremely useful tool that allows fluorescently labelled cells to be visualized in whole-mount preparations. This is particularly advantageous, for example, in studying the dendritic trees of neurons with respect to their environment. One of the most popular, and easiest, ways to visualize a cell is to inject it intracellularly with the fluorophore Lucifer Yellow (LY). However, the argon gas lasers of most LSCM's are not well matched to the excitation spectrum of aqueous LY. When this largely inappropriate excitation is combined with standard filters, designed for fluorescein fluorescence rather than Lucifer Yellow, the resulting image is poor. We report that clearing LY-injected neurons in methyl salicylate and mounting them in Entellan, a non-aqueous medium of high refractive index, enhances their visualization on a Bio-Rad LSCM with standard fluorescein (FITC) filters to an unexpected degree. This technique also leads to a substantial reduction in photobleaching.

Multidentate chelating agents derived from Quinoline-8-aldehyde. II. Stereochemistry of complexes of a tridentate and a sexadentate ligand

Complexes of the sexadentate chelating agent, 1,8-bis(8- quinolylmethyleneimino)-3,6-diazaoctane with iron(II), cobalt(III), and nickel(II) are described. The mode of coordination of this base is such that its complexes must be very closely related structurally to octahedral complexes of 1-amino-2-(8-quinolylmethyleneimino)ethane, and the similarities in the properties of corresponding complexes of both bases reflect this relationship. The resolution of the cobalt(III) complexes of both bases into their stable enantiomorphous forms is described.

Multidentate chelating agents from Quinoline-8-aldehyde. I. Tridentate and quadridentate Schiff bases from primary diamines

Complexes of 1,2-bis(8-quinolylmethyleneimino)ethane with cobalt(II), nickel(II), and copper(II) are described. The copper complexes are five-covalent and, it is suggested, trigonal bipyramidal. Reaction of the Schiff base with iron(II), cobalt(III), and, under certain conditions, nickel(II) leads to its partial hydrolysis, and complexes of the tridentate 1-amino-2-(8-quinolylmethyleneimino)ethane are obtained. The hydrolysis of the Schiff base in the presence of metal ions is discussed in terms of steric effects within the base and of the stereochemical demands of the metal ions. Complexes of 1,2-bis(8-quinolylmethyleneimino)benzene and its hydrolysis derivative 1-amino-2-(8-quinolylmethyleneimino)ethane are also described.