SLEMM: million-scale genomic predictions with window-based SNP weighting.

MOTIVATION: The amount of genomic data is increasing exponentially. Using many genotyped and phenotyped individuals for genomic prediction is appealing yet challenging. RESULTS: We present SLEMM (short for Stochastic-Lanczos-Expedited Mixed Models), a new software tool, to address the computational challenge. SLEMM builds on an efficient implementation of the stochastic Lanczos algorithm for REML in a framework of mixed models. We further implement SNP weighting in SLEMM to improve its predictions. Extensive analyses on seven public datasets, covering 19 polygenic traits in three plant and three livestock species, showed that SLEMM with SNP weighting had overall the best predictive ability among a variety of genomic prediction methods including GCTA's empirical BLUP, BayesR, KAML, and LDAK's BOLT and BayesR models. We also compared the methods using nine dairy traits of ∼300k genotyped cows. All had overall similar prediction accuracies, except that KAML failed to process the data. Additional simulation analyses on up to 3 million individuals and 1 million SNPs showed that SLEMM was advantageous over counterparts as for computational performance. Overall, SLEMM can do million-scale genomic predictions with an accuracy comparable to BayesR. AVAILABILITY AND IMPLEMENTATION: The software is available at https://github.com/jiang18/slemm.

Imputation accuracy from low- to medium-density SNP chips for US crossbred dairy cattle.

This study aimed at evaluating the quality of imputation accuracy (IA) by marker (IAm) and by individual (IAi) in US crossbred dairy cattle. Holstein × Jersey crossbreds were used to evaluate IA from a low- (7K) to a medium-density (50K) SNP chip. Crossbred animals, as well as their sires (53), dams (77), and maternal grandsires (63), were all genotyped with a 78K SNP chip. Seven different scenarios of reference populations were tested, in which some scenarios used different family relationships and others added random unrelated purebred and crossbred individuals to those different family relationship scenarios. The same scenarios were tested on Holstein and Jersey purebred animals to compare these outcomes against those attained in crossbred animals. The genotype imputation was performed with findhap (version 4) software (VanRaden, 2015). There were no significant differences in IA results depending on whether the sire of imputed individuals was Holstein and the dam was Jersey, or vice versa. The IA increased significantly with the addition of related individuals in the reference population, from 86.70 ± 0.06% when only sires or dams were included in the reference population to 90.09 ± 0.06% when sire (S), dam (D), and maternal grandsire genomic data were combined in the reference population. In all scenarios including related individuals in the reference population, IAm and IAi were significantly superior in purebred Jersey and Holstein animals than in crossbreds, ranging from 90.75 ± 0.06 to 94.02 ± 0.06%, and from 90.88 ± 0.11 to 94.04 ± 0.10%, respectively. Additionally, a scenario called SPB+DLD(where PB indicates purebread and LD indicates low density), similar to the genomic evaluations performed on US crossbred dairy, was tested. In this scenario, the information from the 5 evaluated breeds (Ayrshire, Brown Swiss, Guernsey, Holstein, and Jersey) genotyped with a 50K SNP chip and genomic information from the dams genotyped with a 7K SNP chip were combined in the reference population, and the IAm and IAi were 80.87 ± 0.06% and 80.85 ± 0.08%, respectively. Adding randomly nonrelated genotyped individuals in the reference population reduced IA for both purebred and crossbred cows, except for scenario SPB+DLD, where adding crossbreds to the reference population increased IA values. Our findings demonstrate that IA for US Holstein × Jersey crossbred ranged from 85 to 90%, and emphasize the significance of designing and defining the reference population for improved IA.

Inbreeding depression for producer-recorded udder, metabolic, and reproductive diseases in US dairy cattle.

This study leveraged a growing dataset of producer-recorded phenotypes for mastitis, reproductive diseases (metritis and retained placenta), and metabolic diseases (ketosis, milk fever, and displaced abomasum) to investigate the potential presence of inbreeding depression for these disease traits. Phenotypic, pedigree, and genomic information were obtained for 354,043 and 68,292 US Holstein and Jersey cows, respectively. Total inbreeding coefficients were calculated using both pedigree and genomic information; the latter included inbreeding estimates obtained using a genomic relationship matrix and runs of homozygosity. We also generated inbreeding coefficients based on the generational inbreeding for recent and old pedigree inbreeding, for different run-of-homozygosity length classes, and for recent and old homozygous-by-descent segment-based inbreeding. Estimates on the liability scale revealed significant evidence of inbreeding depression for reproductive-disease traits, with an increase in total pedigree and genomic inbreeding showing a notable effect for recent inbreeding. However, we found inconsistent evidence for inbreeding depression for mastitis or any metabolic diseases. Notably, in Holsteins, the probability of developing displaced abomasum decreased with inbreeding, particularly for older inbreeding. Estimates of disease probability for cows with low, average, and high inbreeding levels did not significantly differ across any inbreeding coefficient and trait combination, indicating that although inbreeding may affect disease incidence, it likely plays a smaller role compared with management and environmental factors.

Dry matter intake in US Holstein cows: exploring the genomic and phenotypic impact of milk components and body weight composite.

Large data sets allow estimating feed required for individual milk components or body maintenance. Phenotypic regressions are useful for nutrition management, but genetic regressions are more useful in breeding programs. Dry matter intake (DMI) records from 8,513 lactations of 6,621 Holstein cows were predicted from phenotypes or genomic evaluations for milk components and body size traits. The mixed models also included days in milk, age-parity subclass, trial date, management group, and body weight change during 28- and 42-d feeding trials in mid-lactation. Phenotypic regressions of DMI on milk (0.014 ± 0.006), fat (3.06 ± 0.01), and protein (4.79 ± 0.25) were much less than corresponding genomic regressions (0.08 ± 0.03, 11.30 ± 0.47, and 9.35 ± 0.87) or sire genomic regressions multiplied by 2 (0.048 ± 0.04, 6.73 ± 0.94, and 4.98 ± 1.75). Thus, marginal feed costs as fractions of marginal milk revenue were higher from genetic than phenotypic regressions. According to the energy-corrected milk formula, fat production requires 69% more DMI than protein production. In the phenotypic regression, it was estimated that protein production requires 56% more DMI than fat. However, the genomic regression for the animal showed a difference of only 21% more DMI for protein compared with fat, while the sire genomic regressions indicated approximately 35% more DMI for fat than protein. Estimates of annual maintenance in kg DMI / kg body weight/lactation were similar from phenotypic regression (5.9 ± 0.14), genomic regression (5.8 ± 0.31), and sire genomic regression multiplied by 2 (5.3 ± 0.55) and are larger than those estimated by NASEM (2021) based on NEL equations. Multiple regressions on genomic evaluations for the 5 type traits in body weight composite (BWC) showed that strength was the type trait most associated with body weight and DMI, agreeing with the current BWC formula, whereas other traits were less useful predictors, especially for DMI. The Net Merit formula used to weight different genetic traits to achieve an economically optimal overall selection response was revised in 2021 to better account for these estimated regressions. To improve profitability, breeding programs should select smaller cows with negative residual feed intake that produce more milk, fat, and protein.

A Million-Cow Validation of a Chromosome 14 Region Interacting with All Chromosomes for Fat Percentage in U.S. Holstein Cows.

A genome-wide association study (GWAS) of fat percentage (FPC) using 1,231,898 first lactation cows and 75,198 SNPs confirmed a previous result that a Chr14 region about 9.38 Mb in size (0.14-9.52 Mb) had significant inter-chromosome additive × additive (A×A) effects with all chromosomes and revealed many new such effects. This study divides this 9.38 Mb region into two sub-regions, Chr14a at 0.14-0.88 Mb (0.74 Mb in size) with 78% and Chr14b at 2.21-9.52 Mb (7.31 Mb in size) with 22% of the 2761 significant A×A effects. These two sub-regions were separated by a 1.3 Mb gap at 0.9-2.2 Mb without significant inter-chromosome A×A effects. The PPP1R16A-FOXH1-CYHR1-TONSL (PFCT) region of Chr14a (29 Kb in size) with four SNPs had the largest number of inter-chromosome A×A effects (1141 pairs) with all chromosomes, including the most significant inter-chromosome A×A effects. The SLC4A4-GC-NPFFR2 (SGN) region of Chr06, known to have highly significant additive effects for some production, fertility and health traits, specifically interacted with the PFCT region and a Chr14a region with CPSF1, ADCK5, SLC52A2, DGAT1, SMPD5 and PARP10 (CASDSP) known to have highly significant additive effects for milk production traits. The most significant effects were between an SNP in SGN and four SNPs in PFCT. The CASDSP region mostly interacted with the SGN region. In the Chr14b region, the 2.28-2.42 Mb region (138.46 Kb in size) lacking coding genes had the largest cluster of A×A effects, interacting with seventeen chromosomes. The results from this study provide high-confidence evidence towards the understanding of the genetic mechanism of FPC in Holstein cows.

Large-Sample Genome-Wide Association Study of Resistance to Retained Placenta in U.S. Holstein Cows.

A genome-wide association study of resistance to retained placenta (RETP) using 632,212 Holstein cows and 74,747 SNPs identified 200 additive effects with p-values < 10-8 on thirteen chromosomes but no dominance effect was statistically significant. The regions of 87.61-88.74 Mb of Chr09 about 1.13 Mb in size had the most significant effect in LOC112448080 and other highly significant effects in CCDC170 and ESR1, and in or near RMND1 and AKAP12. Four non-ESR1 genes in this region were reported to be involved in ESR1 fusions in humans. Chr23 had the largest number of significant effects that peaked in SLC17A1, which was involved in urate metabolism and transport that could contribute to kidney disease. The PKHD1 gene contained seven significant effects and was downstream of another six significant effects. The ACOT13 gene also had a highly significant effect. Both PKHD1 and ACOT13 were associated with kidney disease. Another highly significant effect was upstream of BOLA-DQA2. The KITLG gene of Chr05 that acts in utero in germ cell and neural cell development, and hematopoiesis was upstream of a highly significant effect, contained a significant effect, and was between another two significant effects. The results of this study provided a new understanding of genetic factors underlying RETP in U.S. Holstein cows.

Comprehensive analyses of 723 transcriptomes enhance genetic and biological interpretations for complex traits in cattle.

By uniformly analyzing 723 RNA-seq data from 91 tissues and cell types, we built a comprehensive gene atlas and studied tissue specificity of genes in cattle. We demonstrated that tissue-specific genes significantly reflected the tissue-relevant biology, showing distinct promoter methylation and evolution patterns (e.g., brain-specific genes evolve slowest, whereas testis-specific genes evolve fastest). Through integrative analyses of those tissue-specific genes with large-scale genome-wide association studies, we detected relevant tissues/cell types and candidate genes for 45 economically important traits in cattle, including blood/immune system (e.g., CCDC88C) for male fertility, brain (e.g., TRIM46 and RAB6A) for milk production, and multiple growth-related tissues (e.g., FGF6 and CCND2) for body conformation. We validated these findings by using epigenomic data across major somatic tissues and sperm. Collectively, our findings provided novel insights into the genetic and biological mechanisms underlying complex traits in cattle, and our transcriptome atlas can serve as a primary source for biological interpretation, functional validation, studies of adaptive evolution, and genomic improvement in livestock.

Improving national fertility evaluations by accounting for the rapid rise of embryo transfer in US dairy cattle.

Dairy producers have improved fertility of their herds by selecting bulls with higher conception rate evaluations. This research was motivated by the rapid increase in embryo transfer (ET) use to 11% of recent births and >1 million total births, with >5 times as many ET calves born in the United States in 2021 compared with just 5 yr earlier. Historical data used in genetic evaluations are stored in the National Cooperator Database. Recent records in the national pedigree database revealed that only 1% of ET calves have corresponding ET records in the breeding event database, 2% are incorrectly reported as artificial inseminations, and 97% have no associated breeding event. Embryo donation events are also rarely reported. Herd years reporting >10% of calves born by ET but less than half of the expected number of ET breeding events were removed to avoid potential biases. Heifer, cow, and sire conception rate evaluations were recalculated with this new data set according to the methods used for the official national evaluations. The edits removed about 1% of fertility records in the most recent 4 yr. Subsequent analysis showed that censoring herd years with inconsistent ET reporting had little effect on most bulls except for the highest ranking, younger bulls popular for ET use, and with largest effects on genomic selection. Improved ET reporting will be critical for providing accurate fertility evaluations, especially as the popularity of these advanced reproductive technologies continues to rise.

Invited review: Reliability computation from the animal model era to the single-step genomic model era.

The calculation of exact reliabilities involving the inversion of mixed model equations poses a heavy computational challenge when the system of equations is large. This has prompted the development of different approximation methods. We give an overview of the various methods and computational approaches in calculating reliability from the era before the animal model to the era of single-step genomic models. The different methods are discussed in terms of modeling, development, and applicability in large dairy cattle populations. The paper also describes the problems faced in reliability computation. Many details dispersed throughout the literature are presented in this paper. It is clear that a universal solution applicable to every model and input data may not be possible, but we point out several efficient and accurate algorithms developed recently for a variety of very large genomic evaluations.

A Million-Cow Genome-Wide Association Study of Three Fertility Traits in U.S. Holstein Cows.

A genome-wide association study (GWAS) of the daughter pregnancy rate (DPR), cow conception rate (CCR), and heifer conception rate (HCR) using 1,001,374-1,194,736 first-lactation Holstein cows and 75,140-75,295 SNPs identified 7567, 3798, and 726 additive effects, as well as 22, 27, and 25 dominance effects for DPR, CCR, and HCR, respectively, with log10(1/p) > 8. Most of these effects were new effects, and some new effects were in or near genes known to affect reproduction including GNRHR, SHBG, and ESR1, and a gene cluster of pregnancy-associated glycoproteins. The confirmed effects included those in or near the SLC4A4-GC-NPFFR2 and AFF1 regions of Chr06 and the KALRN region of Chr01. Eleven SNPs in the CEBPG-PEPD-CHST8 region of Chr18, the AFF1-KLHL8 region of Chr06, and the CCDC14-KALRN region of Chr01 with sharply negative allelic effects and dominance values for the recessive homozygous genotypes were recommended for heifer culling. Two SNPs in and near the AGMO region of Chr04 that were sharply negative for HCR and age at first calving, but slightly positive for the yield traits could also be considered for heifer culling. The results from this study provided new evidence and understanding about the genetic variants and genome regions affecting the three fertility traits in U.S. Holstein cows.

Invited review: Unknown-parent groups and metafounders in single-step genomic BLUP.

Single-step genomic BLUP (ssGBLUP) is a method for genomic prediction that integrates matrices of pedigree (A) and genomic (G) relationships into a single unified additive relationship matrix whose inverse is incorporated into a set of mixed model equations (MME) to compute genomic predictions. Pedigree information in dairy cattle is often incomplete. Missing pedigree potentially causes biases and inflation in genomic estimated breeding values (GEBV) obtained with ssGBLUP. Three major issues are associated with missing pedigree in ssGBLUP, namely biased predictions by selection, missing inbreeding in pedigree relationships, and incompatibility between G and A in level and scale. These issues can be solved using a proper model for unknown-parent groups (UPG). The theory behind the use of UPG is well established for pedigree BLUP, but not for ssGBLUP. This study reviews the development of the UPG model in pedigree BLUP, the properties of UPG models in ssGBLUP, and the effect of UPG on genetic trends and genomic predictions. Similarities and differences between UPG and metafounder (MF) models, a generalized UPG model, are also reviewed. A UPG model (QP) derived using a transformation of the MME has a good convergence behavior. However, with insufficient data, the QP model may yield biased genetic trends and may underestimate UPG. The QP model can be altered by removing the genomic relationships linking GEBV and UPG effects from MME. This altered QP model exhibits less bias in genetic trends and less inflation in genomic predictions than the QP model, especially with large data sets. Recently, a new model, which encapsulates the UPG equations into the pedigree relationships for genotyped animals, was proposed in simulated purebred populations. The MF model is a comprehensive solution to the missing pedigree issue. This model can be a choice for multibreed or crossbred evaluations if the data set allows the estimation of a reasonable relationship matrix for MF. Missing pedigree influences genetic trends, but its effect on the predictability of genetic merit for genotyped animals should be negligible when many proven bulls are genotyped. The SNP effects can be back-solved using GEBV from older genotyped animals, and these predicted SNP effects can be used to calculate GEBV for young-genotyped animals with missing parents.

Genomic characterization of autozygosity and recent inbreeding trends in all major breeds of US dairy cattle.

Maintaining a genetically diverse dairy cattle population is critical to preserving adaptability to future breeding goals and avoiding declines in fitness. This study characterized the genomic landscape of autozygosity and assessed trends in genetic diversity in 5 breeds of US dairy cattle. We analyzed a sizable genomic data set containing 4,173,679 pedigreed and genotyped animals of the Ayrshire, Brown Swiss, Guernsey, Holstein, and Jersey breeds. Runs of homozygosity (ROH) of 2 Mb or longer in length were identified in each animal. The within-breed means for number and the combined length of ROH were highest in Jerseys (62.66 ± 8.29 ROH and 426.24 ± 83.40 Mb, respectively; mean ± SD) and lowest in Ayrshires (37.24 ± 8.27 ROH and 265.05 ± 85.00 Mb, respectively). Short ROH were the most abundant, but moderate to large ROH made up the largest proportion of genome autozygosity in all breeds. In addition, we identified ROH islands in each breed. This revealed selection patterns for milk production, productive life, health, and reproduction in most breeds and evidence for parallel selective pressure for loci on chromosome 6 between Ayrshire and Brown Swiss and for loci on chromosome 20 between Holstein and Jersey. We calculated inbreeding coefficients using 3 different approaches, pedigree-based (FPED), marker-based using a genomic relationship matrix (FGRM), and segment-based using ROH (FROH). The average inbreeding coefficient ranged from 0.06 in Ayrshires and Brown Swiss to 0.08 in Jerseys and Holsteins using FPED, from 0.22 in Holsteins to 0.29 in Guernsey and Jerseys using FGRM, and from 0.11 in Ayrshires to 0.17 in Jerseys using FROH. In addition, the effective population size at past generations (5-100 generations ago), the yearly rate of inbreeding, and the effective population size in 3 recent periods (2000-2009, 2010-2014, and 2015-2018) were determined in each breed to ascertain current and historical trends of genetic diversity. We found a historical trend of decreasing effective population size in the last 100 generations in all breeds and breed differences in the effect of the recent implementation of genomic selection on inbreeding accumulation.

Gamevar.f90: a software package for calculating individual gametic diversity.

BACKGROUND: Traditional selection in livestock and crops focuses on additive genetic values or breeding values of the individuals. While traditional selection utilizes variation between individuals, differences between gametes within individuals have been less frequently exploited in selection programs. With the successful implementation of genomic selection in livestock and crops, estimation and selection for gametic variation is becoming possible. RESULTS: The gamevar.f90 software is designed to estimate individual-level variance of genetic values of gametes for complex traits in large populations. The software estimates the (co)variances of gametic diversity as well as other diversity parameters that are useful for selection programs and mating designs. The calculation is carried out chromosome by chromosome and can be easily parallelized. The gamevar.f90 program is written in Fortran with efficient computing algorithms in a user-friendly software package with easily-handled input and output files. Finally, we applied the program to estimate gametic variance for hundreds of bulls for lifetime net merit, productive life, and livability. The RPTA (relative predicted transmitting ability), assuming a future selection intensity (if) of 1.5, showed larger variance than GEBV/2, indicating that greater future genetic gains can be obtained using an index that includes gametic variances. We also used the relative coefficient of variation to estimate with 95% confidence the sample sizes required to observe 90% variability of the progeny for lifetime net merit (or to allow at maximum 10% of change in the EBV predicted from progeny data). CONCLUSIONS: Collectively, we develop an efficient computer program package, gamevar.f90, for estimating gametic variance for large numbers of individuals. The novel information on gametic variation will be useful in future animal and crop breeding programs.

GWAS and fine-mapping of livability and six disease traits in Holstein cattle.

BACKGROUND: Health traits are of significant economic importance to the dairy industry due to their effects on milk production and associated treatment costs. Genome-wide association studies (GWAS) provide a means to identify associated genomic variants and thus reveal insights into the genetic architecture of complex traits and diseases. The objective of this study is to investigate the genetic basis of seven health traits in dairy cattle and to identify potential candidate genes associated with cattle health using GWAS, fine mapping, and analyses of multi-tissue transcriptome data. RESULTS: We studied cow livability and six direct disease traits, mastitis, ketosis, hypocalcemia, displaced abomasum, metritis, and retained placenta, using de-regressed breeding values and more than three million imputed DNA sequence variants. After data edits and filtering on reliability, the number of bulls included in the analyses ranged from 11,880 (hypocalcemia) to 24,699 (livability). GWAS was performed using a mixed-model association test, and a Bayesian fine-mapping procedure was conducted to calculate a posterior probability of causality to each variant and gene in the candidate regions. The GWAS detected a total of eight genome-wide significant associations for three traits, cow livability, ketosis, and hypocalcemia, including the bovine Major Histocompatibility Complex (MHC) region associated with livability. Our fine-mapping of associated regions reported 20 candidate genes with the highest posterior probabilities of causality for cattle health. Combined with transcriptome data across multiple tissues in cattle, we further exploited these candidate genes to identify specific expression patterns in disease-related tissues and relevant biological explanations such as the expression of Group-specific Component (GC) in the liver and association with mastitis as well as the Coiled-Coil Domain Containing 88C (CCDC88C) expression in CD8 cells and association with cow livability. CONCLUSIONS: Collectively, our analyses report six significant associations and 20 candidate genes of cattle health. With the integration of multi-tissue transcriptome data, our results provide useful information for future functional studies and better understanding of the biological relationship between genetics and disease susceptibility in cattle.

Changes in genetic selection differentials and generation intervals in US Holstein dairy cattle as a result of genomic selection.

Seven years after the introduction of genomic selection in the United States, it is now possible to evaluate the impact of this technology on the population. Selection differential(s) (SD) and generation interval(s) (GI) were characterized in a four-path selection model that included sire(s) of bulls (SB), sire(s) of cows (SC), dam(s) of bulls (DB), and dam(s) of cows (DC). Changes in SD over time were estimated for milk, fat, and protein yield; somatic cell score (SCS); productive life (PL); and daughter pregnancy rate (DPR) for the Holstein breed. In the period following implementation of genomic selection, dramatic reductions were seen in GI, especially the SB and SC paths. The SB GI reduced from ∼7 y to less than 2.5 y, and the DB GI fell from about 4 y to nearly 2.5 y. SD were relatively stable for yield traits, although modest gains were noted in recent years. The most dramatic response to genomic selection was observed for the lowly heritable traits DPR, PL, and SCS. Genetic trends changed from close to zero to large and favorable, resulting in rapid genetic improvement in fertility, lifespan, and health in a breed where these traits eroded over time. These results clearly demonstrate the positive impact of genomic selection in US dairy cattle, even though this technology has only been in use for a short time. Based on the four-path selection model, rates of genetic gain per year increased from ∼50-100% for yield traits and from threefold to fourfold for lowly heritable traits.

Ranking sires using genetic selection indices based on financial investment methods versus lifetime net merit.

Current USDA selection indices such as lifetime net merit (NM$) estimate lifetime profit differences, which are accurately approximated by a linear combination of 13 traits. In these indices, every animal gets credit for 2.78 lactations of the traits expressed per lactation, such as fat and protein, independent of its productive life (PL). This formulation may over- or underestimate the net revenue from traits expressed per lactation depending on PL. The objectives were to develop 2 genetic selection indices using financial investment methods to account for differences in PL and to compare them with the 2017 NM$ for marketed Holstein sires. Selection among animals with different PL is an example of investment in mutually exclusive projects that have unequal duration. Financial investment theory says that such projects are best compared with the annualized net present value (ANPV) method when replacement occurs with technologically equal assets. However, genetic progress implies that future available replacement animals are technologically improved assets. Asset replacement theory with improved assets results in an annualized value including genetic opportunity cost (AVOC) for each animal. We developed the ANPV and AVOC and compared these with the NM$ for 1,500 marketed Holstein sires from the December 2017 genetic evaluation. The lowest Pearson correlation coefficient was 0.980 between AVOC and NM$, whereas the highest was 0.999 between ANPV and NM$ among the 1,500 sires. Correlations for the top 300 sires were lower. Although we found high correlations between indices, the 95th and 5th percentiles of individual rank changes between AVOC and NM$ were +131 and -163 positions, respectively, whereas these changes between ANPV and NM$ were +27 and -45 positions, respectively. The relative emphasis of PL in the AVOC index was half of the relative emphasis in NM$. These results show that applying financial investment methods to value differences in genetic merit of animals changes their rankings compared with the NM$ formulation. Rank changes were meaningful enough that the new indices warrant consideration for use in practice.

Cattle Sex-Specific Recombination and Genetic Control from a Large Pedigree Analysis.

Meiotic recombination is an essential biological process that generates genetic diversity and ensures proper segregation of chromosomes during meiosis. From a large USDA dairy cattle pedigree with over half a million genotyped animals, we extracted 186,927 three-generation families, identified over 8.5 million maternal and paternal recombination events, and constructed sex-specific recombination maps for 59,309 autosomal SNPs. The recombination map spans for 25.5 Morgans in males and 23.2 Morgans in females, for a total studied region of 2,516 Mb (986 kb/cM in males and 1,085 kb/cM in females). The male map is 10% longer than the female map and the sex difference is most pronounced in the subtelomeric regions. We identified 1,792 male and 1,885 female putative recombination hotspots, with 720 hotspots shared between sexes. These hotspots encompass 3% of the genome but account for 25% of the genome-wide recombination events in both sexes. During the past forty years, males showed a decreasing trend in recombination rate that coincided with the artificial selection for milk production. Sex-specific GWAS analyses identified PRDM9 and CPLX1 to have significant effects on genome-wide recombination rate in both sexes. Two novel loci, NEK9 and REC114, were associated with recombination rate in both sexes, whereas three loci, MSH4, SMC3 and CEP55, affected recombination rate in females only. Among the multiple PRDM9 paralogues on the bovine genome, our GWAS of recombination hotspot usage together with linkage analysis identified the PRDM9 paralogue on chromosome 1 to be associated in the U.S. Holstein data. Given the largest sample size ever reported for such studies, our results reveal new insights into the understanding of cattle and mammalian recombination.

Dissection of additive, dominance, and imprinting effects for production and reproduction traits in Holstein cattle.

BACKGROUND: Although genome-wide association and genomic selection studies have primarily focused on additive effects, dominance and imprinting effects play an important role in mammalian biology and development. The degree to which these non-additive genetic effects contribute to phenotypic variation and whether QTL acting in a non-additive manner can be detected in genetic association studies remain controversial. RESULTS: To empirically answer these questions, we analyzed a large cattle dataset that consisted of 42,701 genotyped Holstein cows with genotyped parents and phenotypic records for eight production and reproduction traits. SNP genotypes were phased in pedigree to determine the parent-of-origin of alleles, and a three-component GREML was applied to obtain variance decomposition for additive, dominance, and imprinting effects. The results showed a significant non-zero contribution from dominance to production traits but not to reproduction traits. Imprinting effects significantly contributed to both production and reproduction traits. Interestingly, imprinting effects contributed more to reproduction traits than to production traits. Using GWAS and imputation-based fine-mapping analyses, we identified and validated a dominance association signal with milk yield near RUNX2, a candidate gene that has been associated with milk production in mice. When adding non-additive effects into the prediction models, however, we observed little or no increase in prediction accuracy for the eight traits analyzed. CONCLUSIONS: Collectively, our results suggested that non-additive effects contributed a non-negligible amount (more for reproduction traits) to the total genetic variance of complex traits in cattle, and detection of QTLs with non-additive effect is possible in GWAS using a large dataset.

Selecting sequence variants to improve genomic predictions for dairy cattle.

BACKGROUND: Millions of genetic variants have been identified by population-scale sequencing projects, but subsets of these variants are needed for routine genomic predictions or genotyping arrays. Methods for selecting sequence variants were compared using simulated sequence genotypes and real July 2015 data from the 1000 Bull Genomes Project. METHODS: Candidate sequence variants for 444 Holstein animals were combined with high-density (HD) imputed genotypes for 26,970 progeny-tested Holstein bulls. Test 1 included single nucleotide polymorphisms (SNPs) for 481,904 candidate sequence variants. Test 2 also included 249,966 insertions-deletions (InDels). After merging sequence variants with 312,614 HD SNPs and editing steps, Tests 1 and 2 included 762,588 and 1,003,453 variants, respectively. Imputation quality from findhap software was assessed with 404 of the sequenced animals in the reference population and 40 randomly chosen animals for validation. Their sequence genotypes were reduced to the subset of genotypes that were in common with HD genotypes and then imputed back to sequence. Predictions were tested for 33 traits using 2015 data of 3983 US validation bulls with daughters that were first phenotyped after August 2011. RESULTS: The average percentage of correctly imputed variants across all chromosomes was 97.2 for Test 1 and 97.0 for Test 2. Total time required to prepare, edit, impute, and estimate the effects of sequence variants for 27,235 bulls was about 1 week using less than 33 threads. Many sequence variants had larger estimated effects than nearby HD SNPs, but prediction reliability improved only by 0.6 percentage points in Test 1 when sequence SNPs were added to HD SNPs and by 0.4 percentage points in Test 2 when sequence SNPs and InDels were included. However, selecting the 16,648 candidate SNPs with the largest estimated effects and adding them to the 60,671 SNPs used in routine evaluations improved reliabilities by 2.7 percentage points. CONCLUSIONS: Reliabilities for genomic predictions improved when selected sequence variants were added; gains were similar for simulated and real data for the same population, and larger than previous gains obtained by adding HD SNPs. With many genotyped animals, many data sources, and millions of variants, computing strategies must efficiently balance costs of imputation, selection, and prediction to obtain subsets of markers that provide the highest accuracy.

Identification of a nonsense mutation in APAF1 that is likely causal for a decrease in reproductive efficiency in Holstein dairy cattle.

The HH1 haplotype on chromosome 5 is associated with a reduced conception rate and a deficit of homozygotes at the population level in Holstein cattle. The source HH1 haplotype was traced to the bull Pawnee Farm Arlinda Chief (Chief), who was born in 1962 and has sired more than 16,000 daughters. We identified a nonsense mutation in APAF1 (apoptotic protease activating factor 1;APAF1 p.Q579X) within HH1 using whole-genome resequencing of Chief and 3 of his sons. This mutation is predicted to truncate 670 AA (53.7%) of the encoded APAF1 protein that contains a WD40 domain critical to protein-protein interactions. Initial screening revealed no homozygous individuals for the mutation in 758 animals previously genotyped, whereas all 497 HH1 carriers possessed 1 copy of the mutant allele. Subsequent commercial genotyping of 246,773 Holsteins revealed 5,299 APAF1 heterozygotes and zero homozygotes for the mutation. The causative role of this mutation is also supported by functional data in mice that have demonstrated Apaf1 to be an essential molecule in the cytochrome-c-mediated apoptotic cascade and directly implicated in developmental and neurodegenerative disorders. In addition, most Apaf1 homozygous knockouts die by day 16.5 of development. We thus propose that the APAF1 p.Q579X nonsense mutation is the functional equivalent of the Apaf1 knockout. This mutation has caused an estimated 525,000 spontaneous abortions worldwide over the past 35 years, accounting for approximately $420 million in losses. With the mutation identified, selection against the deleterious allele in breeding schemes has aided in eliminating this defect from the population, reducing carrier frequency from 8% in past decades to 2% in 2015.