Source: MICHIGAN STATE UNIV submitted to NRP
HIERARCHICAL MODELING STRATEGIES FOR IMPROVING DAIRY CATTLE PRODUCTION USING GENOMICS AND PHENOMICS DATA
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
COMPLETE
Funding Source
HATCH
Reporting Frequency
Annual
Accession No.
1011789
Grant No.
(N/A)
Cumulative Award Amt.
(N/A)
Proposal No.
(N/A)
Multistate No.
(N/A)
Project Start Date
Mar 1, 2017
Project End Date
Feb 28, 2022
Grant Year
(N/A)
Program Code
[(N/A)]- (N/A)
Recipient Organization
MICHIGAN STATE UNIV
(N/A)
EAST LANSING,MI 48824
Performing Department
ANIMAL SCIENCE
Non Technical Summary
Dairy cattle production has been characterized by steady increases in milk production such that the United States (US) has transformed from a net importer to a net exporter of milk just over the past decade. But with recently depressed milk prices and profit margins, it is vitally important advance foundational research that will lead to the development of more balanced strategies for improving the economic efficiency of US dairy farms. These strategies need to consider a greater emphasis on novel traits characterizing dairy cattle fitness, fertility, and feed efficiency than what has been considered previously in order to ensure the economic and environmental sustainability of the US dairy industry. Already many other countries in Europe and Canada are somewhat ahead in implementing dairy genetic evaluation systems for these novel traits, thereby creating an even greater sense of urgency to spearhead such efforts within the US. However, it can be rather challenging and expensive to collect a meaningfully large data resource on novel traits such that extensive, including international, collaboration is necessary. This project is based on the use of two very broad international efforts used to leverage data resources from outside the US in order to eventually improve fitness, fertility, and feed efficiency in US dairy cattle.The first data resource is the dairy cattle feed efficiency project funded by a USDA-NIFA competitive grant involving feed intake data on nearly 7000 genotyped cows from 16 different research stations in 4 countries, with most of the data originating from within the US. Although classical quantitative genetic analyses on feed efficiency from this dataset have been already conducted, there is a greater need to better model the heterogeneous relationships between the component traits of feed efficiency across these various stations. This also includes identifying chromosomal regions whose genetic control on feed efficiency is environmentally sensitive to, for example, temperature or herd management. Our proposed research involves developing the statistical genetic tools, specifically mixed model or hierarchical Bayesian inference, that can be used to allow these finer inferences in order to facilitate precision genetic management of dairy cattle for feed efficiency in the future.Our second data resource involves the use of milk spectral data already collected from three research farms (8000+ records on 3000+ cows) and commercial dairy herds (5 million+ cows on 800,000+cows) from within the US, Canada, and Italy, with only a very small component of this data deriving from within the US. Spectral data, generated by mid-infrared spectroscopy, is related to the absorption of electromagnetic waves and each wavelength is specific to a particular chemical feature within a molecule. The spectra of milk samples has been demonstrated to infer very specific nutrients or metabolites such as fatty acid profiles in research studies. These fatty acid profiles, in turn, provide physiological "fingerprints" on the cows, including their energy balance and subsequent likelihood of success for various health and fertility outcomes (e.g., pregnancy). Although our Italian and Canadian collaborators have already made extensive use of their data for such purposes, our group at MSU has recently demonstrated that the use of hierarchical Bayesian techniques lead to a more efficient use of the wavelength data to predict fatty acid profiles than the use of the current default analysis method, partial least squares. In order to assess the potential benefit of these developments for US dairy herds, we have started to create a data pipeline of spectral data on milk samples and health and reproduction outcomes from at least a dozen commercial dairy herds in Michigan and Wisconsin. Prediction equations based on the use of hierarchical Bayesian techniques and developed from this international dataset should help boost greater accuracy of prediction on health and reproduction outcomes than those previously reported and facilitate quicker adoption of the use of milk spectral data as a dairy cattle management tool.
Animal Health Component
10%
Research Effort Categories
Basic
80%
Applied
10%
Developmental
10%
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
3033410108060%
3073410209040%
Knowledge Area
303 - Genetic Improvement of Animals; 307 - Animal Management Systems;

Subject Of Investigation
3410 - Dairy cattle, live animal;

Field Of Science
2090 - Statistics, econometrics, and biometrics; 1080 - Genetics;
Goals / Objectives
Our goals are focused around 4 primary objectives with various sub-objectives listed underneath below.Objective 1: Develop genome wide association (GWA) and genomic selection (GS) models that best reflect the genomic architecture within livestock species1a. Demonstrate and test the utility of hierarchical Bayesian regression models for specificity and sensitivity of detection of genomic regions containing candidate genes.1b. Develop and test hierarchical Bayesian GWA regression models that allow joint analysis of data on genotyped and non-genotyped animalsObjective 2: Develop GWA and GS models that further extend those of Objective 1 allowing for environment-specific genetic inferences in the presence of genotype by environment interaction2a. Extend reaction norm (RN) models to allow for multiple covariates and/or their higher order polynomials or interactions thereof in GWA analyses.2b. Develop GWA inference strategies that can be used to formally test for environmentally sensitive genomic regionsObjective 3: Refine hierarchical Bayesian models for multiple trait inferences for individually and jointly modeling genomics and phenomics data.3a. Demonstrate how multiple trait genomic analyses can differentiate between GWA of DMI versus GWA of feed efficiency.3b. Demonstrate in context of traits such as feed efficiency how multiple trait genomic analyses can infer upon genetic marker- specific pleiotropy.Objective 4: Assess the value of high dimensional phenomics data such as mid infrared spectral data to predict milk fatty acid profiles and health and reproduction outcomes in US dairy cows.4a. Compare Bayesian variable selection techniques to conventional partial least squares methods for the prediction of fatty acid profiles, reproduction and health outcomes.4b. To develop and validate novel methods for the use of longitudinal records of mid infrared spectra for the longitudinal prediction of: novel metabolite phenotypes that are directly related to the energy balance of the cow and to health and reproductive outcomes.4c. Develop a data pipeline for Michigan spectral data so that potential benefits of using spectral data to predict reproduction and health outcomes could be realized within the US.
Project Methods
In all cases, we will use simulation studies to test these various objectives to investigate properties such as accuracy (e.g. correlation between true and estimated genetic merit) and the use of receiver operating characteristic (ROC) curves for assessing GWA performance. ROC curves involve plotting specificity versus false positive rates; the area under those curves (AUC) are often used as metrics to directly compare methods. Simulation studies will be based on the use of actual existing livestock genotype and phenotype data structures. We will use cross-validation strategies to assess the performance of methods in real data.For Objective 1, we will compare conventional GBLUP methods with BayesA and variable selection methods, specifically SSVS, for their specificity and sensitivity for GWA detection through the use of ROC-AUC analyses based on the data and genotype structure of the US subset of the dairy feed efficiency consortium dataset. Furthermore, these comparison will be conducted based on single SNP focused versus windows focused inference based on windows of various sizes, ranging from 0.5 to 3 megabases (MB) in length, or based on window sizes adaptively inferred from linkage disequilibrium. We will immediately follow-up that work with research developments that allow the use of BayesA and SSVS to conduct GWA in populations where only a fraction of the animals are genotyped and assess the impact of how the nature of missing genotypes (i.e., some research stations having all missing genotypes versus all research stations having some missing genotypes) affects GWA inference. Much of this work will be based on extending previous Bayesian GWA work so that the key hyperparameters that determine genetic architecture (i.e. proportion of SNP effects that are non-zero) will be estimated as wellFor Objective 2, we create GWA inferences on SNP markers for environmental sensitivities to each of several different covariates simultaneously. Our first application will be the US feed efficiency database where SNP effects will be modeled as a function of temperature, relative humidity, and average herd performance amongst other measures. GWA inferences will be based on single SNP and windows-based (i.e. 1 MB regions) inferences; yet we will also develop a hypothesis testing strategy for GxE based GWA inferences that has been ignored in all previous GxE GWA research. Our initial GxE applications will be based on GBLUP implementations although we hope to eventually extend our GWA inference to allow for covariate-specific RN variance components and sparser prior specifications based on BayesA/B or SSVS.For Objective 3, we will also use the feed efficiency data consortium dataset to investigate GWA inferences for DMI, ME, and MBW in conjunction with various measures of feed efficiency noting again that DMI is not feed efficiency. We will adapt previous work provided at the animal and environment (or management systems) level to address the heterogeneity in genetic correlations between traits at the gene, animal and systems level in conjunction with the earlier work on heterogeneity in correlations between traits at the cow and systems (herd) level.For Objective 4, we will use data from research farms to derive and validate equations to predict metabolites (FA and protein profiles, as well as markers for metabolic diseases). These equations will be applied to spectra collected at research farms to predict longitudinal trajectories for the key metabolites. We will also use data from commercial farms to derive and validate equations for prediction of robustness outcomes from milk spectra. The first is a two-step approach where the equations will be applied to spectra data from commercial farms to predict metabolites and disease markers, and these predicted markers in turn will be used to derive equations to predict health and reproductive outcomes.

Progress 10/01/19 to 09/30/20

Outputs
Target Audience:Our target audience are fellow dairy scientists and quantitative geneticists. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?One post-doctoral research associate (Piush Khanal) has been trained under this project and has been primarily responsible for the work in Objective #4. How have the results been disseminated to communities of interest?Results have been published in the Journal of Dairy Science What do you plan to do during the next reporting period to accomplish the goals?We hope to pursue work on better modeling the effects of body weight change on feed efficiency using random regression methodologies. Furthermore, we hope to use milk spectral data to prospectively predict reproductive outcomes such as days open or pregnancy rate at first breeding.

Impacts
What was accomplished under these goals? Objective 1: No updates Objective 2: No updates Objective 3: A greater number of dairy economic selection indexes are incorporating a measure of feed efficiency (FE) as a key trait. Definitions of FE traits have ranged from DMI to residual feed intake (RFI), noting that RFI is effectively DMI adjusted for various energy sink traits such as body weight (BW) and milk energy (MILKE). Other definitions of FE fall between these two extremes such as feed saved (FS) which combines RFI and the portion of DMI required to maintain BW. The choice between different FE traits can create confusion as to how to meaningfully compare their heritabilities, estimated breeding values (EBV) and their corresponding reliabilities, and how to differentially incorporate these EBV into selection indexes. If RFI and FS are merely linear functions of DMI, BW, and MILKE with known genetic variances and covariances between these 3 traits, there may be no need to directly compute RFI or FS phenotypes in order to determine their heritabilities, genetic correlations, EBV and their respective reliabilities for individual animals. We demonstrated how the estimated total genetic merit is invariant to the specification of a FE trait within a selection index. That is, economic weights for a selection index involving one particular FE trait readily convert into the economic weights for a selection index involving a different linear function of that FE trait. We used these different specifications of FE to provide insight as to the impact of the degree of missingness (e.g., paucity of DMI relative to milk yield records) on the EBV accuracies of the various derivative FE traits. We particularly highlighted that the generally observed higher EBV accuracies for DMI, then for FS, and lastly for RFI are driven by the greater genetic correlations of DMI with BW and MILKE and of FS with BW. Finally, we advocated a genetic regression approach to deriving FS and RFI recognizing that genetic versus residual relationships between FE component traits may differ substantially from each other. Objective 4: An accurate diagnosis of pregnancy is important for better reproductive management of dairy farms. Changes in milk composition traits with the stage of pregnancy are well documented. The milk infrared spectrum reflects the chemical composition of milk. The objectives of this study were to compare the partial least squares regression and Bayesian models for prediction of pregnant status (PS) from spectral profiles, investigate the potential of whole infrared spectral profiles for prediction of PS at different stages and identify the wavelength regions that are highly associated with different stages of pregnancy. Milk spectral data and pregnant were obtained from Holstein cows on 123 Michigan dairy herds. Data were pretreated with first and second derivative of raw spectra. Seven different stages based on days after insemination were created where pregnant and their matching non-pregnant contemporary herdmates were required to be within the same stage (± 10 d for DIM). Partial least squares (PLS) and Bayesian regression methods with were applied to determine the predictive accuracy (PA) of PS based on 10-fold herd independent cross-validation. Wavelength region associations were determined for each stage using a windows-based approach, where windows were clustered based on correlation among wavelengths. Bayesian regression methods demonstrated higher PA (P < 0.05) compared to PLS at all stages except for non-significant differences at the first two stages (1-30 and 31-60d). The PA at stage 7 (≥180d) was 13% greater compared to stage 1 (1-30d) whereas the PA of stage 2, 3, 4 and 5 were similar and increased by 3% at stage 6 (151-180d). The results showed that pregnancy was highly associated (posterior probability = 1) with regions within the range of wavelengths from 1,063 cm-1 to 1,134 cm-1, 1,201 cm-1 to 1,257 cm-1 and 1,260 cm-1 to 1,432 cm-1 for all stages. Wavelengths ranging of 1,730 cm-1 to 1,764 cm-1, 1,775 cm-1 to 1,992 cm-1, 1,995 cm-1 to 2,167 cm-1 to 2,316 cm-1 had greater association with pregnancy although it varied across stages. The signal of pregnancy in range of wavelengths from 1,477 cm-1 to 1,507 cm-1, 1,510 cm-1 to 1,574 cm-1 was lower for earlier stages compared to later stages whereas for the regions of 2,984 to 3,077 cm-1 and 3,081 cm-1 to 3,133cm-1 the effect of pregnancy was higher for earlier stage. This study provides new insight in determining the important regions associated with pregnancy and helps in screening of pregnant animals.

Publications

  • Type: Journal Articles Status: Published Year Published: 2019 Citation: Li, B, L. Fang, D.J. Null, J.L. Hutchison, E.E. Connor, P.M. VanRaden, M.J. Vandehaar, R.J. Tempelman, K.A. Weigel, and J.B. Cole. 2019. High-density genome-wide association study for residual feed intake in Holstein dairy cattle. Journal of Dairy Science 102:11067-11080.
  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Li, B, P.M. VanRaden, E. Gunak, J.R. OConnell, D.J. Null, E.E. Connor, M.J. Vandehaar, R.J. Tempelman, K.A. Weigel, and J.B. Cole. 2020. Genomic prediction of residual feed intake in US dairy cattle. Journal of Dairy Science 103:2477-2486.
  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Tempelman, R.J. and Y. Lu. 2020. Genetic relationships between different measures of feed efficiency and the implications for dairy cattle selection indexes. Journal of Dairy Science 103: 5327-5345.


Progress 10/01/18 to 09/30/19

Outputs
Target Audience:Our target audience are fellow dairy scientists and quantitative geneticists. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?One post-doctoral research associate (Gabriel Rovere) has been trained under this project and has been primarily responsible for the work in Objective #4. How have the results been disseminated to communities of interest?Yes, the results have been presented in journal articles and at the national meetings of the American Dairy Science Association. What do you plan to do during the next reporting period to accomplish the goals?We hope to publish some key genotype by environment interaction work as it pertains to ambient temperature effects on economically important dairy traits within the coming year. We also hope to infer upon associations between milk spectral data absorbances and milk fatty acids and and fertility traits using Michigan dairy cattle as a follow up to our work with other populations (Canadian and Italian Holsteins) in past years. We also plan to publish a paper delineating the relationships between various feed efficiency traits by genetic dissection of a multiple trait analysis of the component energy sink and source traits.

Impacts
What was accomplished under these goals? Objective 1: We studied genome-wide associations of high-density genotypes with residual feed intake (RFI) in Holstein cattle to identify candidate genes and biological pathways. Data were from 4,823 lactations of 3,947 Holstein cows in research herds in the USA. RFI was shown to be a highly polygenic trait being regulated by many small-sized effects. The top signals for RFI were in genetic regions previously associated with feed intake, energy balance, digestion and metabolism of carbohydrates and proteins, immunity, mitochondrial activities, rumen development, skeletal development, and spermatogenesis. The regions of 40.7 to 41.5 Mb on chromosome 25 and 57.7 to 58.2 Mb on chromosome 18 were the top associated regions for RFI. Objective 2: No updates Objective 3: No updates. Objective 4: Fourier-transform near- and mid-infrared (FTIR) milk spectral data are routinely collected in many countries worldwide. Establishing an optimal strategy to use spectral data in genetic evaluations requires knowledge of the heritabilities of individual FTIR wavelength absorbances. Previous FTIR heritability estimates have been based on relatively small sample sizes and have not considered the possibility that heritability may vary across parities and stages of the lactation. We used data from ∼370,000 test-day records of Canadian Holstein cows to produce a landscape of the heritability of FTIR spectra, 1,060 wavelengths in the near- and mid-infrared spectrum (5,011-925 cm−1), by parity and month of the lactation (mo 1 to 3 and mo 1 to 6, respectively). The 2 regions of the spectrum associated with absorption of electromagnetic energy by water molecules were estimated to have very high phenotypic variances, very low heritabilities, and very low proportion of variance explained by herd-year-season (HYS) subclasses. The near- or short-wavelength infrared (SWIR: 5,066-3,672 cm−1) region was also characterized by low heritability estimates, whereas the estimated proportion of the variance explained by HYS was high. The mid-wavelength infrared region (MWIR: 3,000-2,500 cm−1) and the transition between mid and long-wavelength infrared region (MWIR-LWIR: 1,500-925 cm−1) harbor several waves characterized by moderately high (≥0.4) heritabilities. Most of the high-heritability regions contained wavelengths that are reported to be associated with important milk metabolites and components. Interestingly, these 2 same regions tended to show more variability in heritabilities between parity and lactation stage. Second parity showed heritability patterns that were distinctly different from those of the first and third parities, whereas the first 2 mo of the lactation had clearly distinct heritability patterns compared with mo 3 to 6. As of August 20, 2019, we have already collected 728,118 spectral data records collected during 2019 on over 605 Central Star enrolled herds. Given that this level of data collection can continue to be supported, we should be able to collect spectral data records on over 1 million Michigan milk samples per year! This should put the state of Michigan in the forefront of collecting spectral data records to predict various measures of dairy cattle performance, including reproduction. We hope to cross-reference our fatty acid inferences with spectral data associations involving reproduction data to allow additional follow-up.

Publications

  • Type: Conference Papers and Presentations Status: Other Year Published: 2018 Citation: Tempelman, R.J. 2018. Dry matter intake versus residual feed intake for genetic evaluation of dairy cattle for feed efficiency. Invited Seminar presented at the Efficient Dairy Genome Project, Guelph, Ontario,CANADA, December 10, 2018.
  • Type: Conference Papers and Presentations Status: Other Year Published: 2019 Citation: Tempelman, R.J. 2019. The quantitative genetics of feed efficiency in dairy cattle. Invited Seminar presented at the Chinese Cattle Science Conference, Taiin, China, May 25, 2019.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2019 Citation: Tempelman, R.J. and Y. Lu. 2019. Genetic relationships between different measures of feed efficiency and the implications for dairy cattle selection indexes. Invited paper presented at the American Dairy Science Association Meetings, Cincinatti, OH., June 24, 2019.
  • Type: Journal Articles Status: Published Year Published: 2019 Citation: Rovere, G., G. de los Campos, R. J. Tempelman, A. I. Vazquez, F. Miglior, F. Schenkel, A. Cecchinato, G. Bittante, H. Toledo-Alvarado, and A. Fleming. 2019. A landscape of the heritability of Fourier-Transform infrared spectral wavelengths of milk samples by parity and lactation stage in Holstein cows. Journal of Dairy Science 102:1354-1363.
  • Type: Journal Articles Status: Published Year Published: 2019 Citation: deSouza, R.A., R.J. Tempelman, M.S. Allen, and M.J. Vandehaar. 2019. Updating the prediction of dry matter intake of lactating dairy cows. Journal of Dairy Science 102:7948-7960.
  • Type: Conference Papers and Presentations Status: Other Year Published: 2018 Citation: Tempelman, R.J. 2018. Some insights into the quantitative genetic modeling of feed efficiency. Invited Seminar presented for the Animal Breeding and Genetics Group, Department of Animal Science, Iowa State University, October 16, 2018.
  • Type: Conference Papers and Presentations Status: Other Year Published: 2018 Citation: Tempelman, R.J. 2018. Some insights into the quantitative genetic modeling of feed efficiency and its implications for multiple trait genomic evaluations. Invited Seminar presented at the Genetic Prediction Workshop, Beef Improvement Federation, Kansas City, MO, December 5, 2018.
  • Type: Journal Articles Status: Accepted Year Published: 2019 Citation: Li, B, L. Fang, D.J. Null, J.L. Hutchison, E.E. Connor, P.M. VanRaden, M.J. Vandehaar, R.J. Tempelman, K.A. Weigel, and J.B. Cole. 2019. High-density genome-wide association study for residual feed intake in Holstein dairy cattle. Journal of Dairy Science (in press).


Progress 10/01/17 to 09/30/18

Outputs
Target Audience:Our target audience are fellow dairy scientists and quantitative geneticists Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?One post-doctoral research associate (Gabriel Rovere) has been trained under this project. Dr. Rovere is currently responsible for implementing much of the work for Objective #4 How have the results been disseminated to communities of interest?Yes, the results have been presented in journal articles and at the national meetings of the American Dairy Science Association. What do you plan to do during the next reporting period to accomplish the goals?We hope to publish some key genotype by environment interaction work as it pertains to ambient temperature effects on economically important dairy traits within the coming year. Furthermore, we are not integrating our work on DNA genotyping and spectral data analysis of milk samples to determine the heritability of spectral absorbances at various wavelengths along in the mid-infrared spectrum while at the same time using Bayesian variable techniques to infer which particular wavelength bands are important for various milk fatty acids. That integration may allow us to directly select for wavelength absorbances as proxy for milk fatty acids that are important for nutrition in future genetic selection index programs. We also hope to infer upon associations between milk spectral data absorbances and fertility and health traits in Michigan dairy cattle as a follow up to our work with other populations in the past year.

Impacts
What was accomplished under these goals? The potential impact of this project is that it provides a analytical tools involving high dimensional data to facilitate both dairy management and genetic selection. Large datasets are created based on DNA genotyping of cattle to infer their genetic potential or spectral analysis of milk samples which can be used to provide "physiological fingerprints" phenomic information on cows with respect to their health, fertility and energy balance. The further advancement of such "big data" genetic and phenomic tools will be vital in best harnessing important and expensive to collect novel traits like feed intakes in order to improve feed efficiency (producing more milk for the same amount of feed intake). Objective 1a. Anti-Mullerian hormone (AMH) is an ovarian growth factor that plays an important role in regulation of ovarian follicle growth. Concentrations of AMH were determined in 2,905 dairy Holstein heifers genotyped using the Zoetis medium density panel (Zoetis Inclusions, Kalamazoo, MI) with 54,519 single nucleotide polymorphism (SNP) markers remaining after standard genotype quality control edits. The genomic heritability (+/- standard error of the mean) of AMH was estimated to be 0.36 +/- 0.03. Our GWA analysis inferred significant associations between AMH and 11 SNP markers on chromosome 11 and 1 SNP marker on chromosome 20. Gene set enrichment analysis revealed that 2 gene ontology (GO) terms were significantly enriched in the list of candidate genes: G-protein coupled receptor signaling pathway (GO:0007186) and the detection of chemical stimulus involved in sensory perception (GO:0050907). The estimated high heritability and previously established associations between AMH and ovarian follicular reserve, fertility, longevity, and superovulatory response in cattle implies that AMH could be used as a biomarker for genetic improvement of reproductive potential. Objective 1b. Nothing new to report this year. Objectives 2a and 2b. Nothing new to report this year. Objectives 3a and 3b Genome-wide association (GWA) of feed efficiency (FE) could help target important genomic regions influencing FE. Data provided by an international dairy FE research consortium consisted of phenotypic records on dry matter intakes (DMI), milk energy (MILKE), and metabolic body weight (MBW) on 6,937 cows from 16 stations in 4 counties. Of these cows, 4,916 had genotypes on 57,347 single nucleotide polymorphism (SNP) markers. We compared a GWA analysis based on the more classical residual feed intake (RFI) model with one based on a previously proposed multiple trait (MT) approach for modeling FE using an alternative measure (DMI vertical bar MILKE, MBW). Both models were based on a single-step genomic BLUP procedure that allowed the use of phenotypes from both genotyped and nongenotyped cows. Estimated effects for single SNP markers were small and not statistically important but virtually identical for either FE measure (RFI vs. DMI vertical bar MILKE, MBW). However, upon further refining this analysis to develop joint tests within nonoverlapping 1-Mb windows, significant associations were detected between either measure of FE with a window on each of Bos taurus autosomes BTA12 and BTA26. There was, as expected, no overlap between detected genomic regions for DMI vertical bar MILKE, MBW and genomic regions influencing the energy sink traits (i.e., MILKE and MBW) because of orthogonal relationships clearly defined between the various traits. Conversely, GWA inferences on DMI can be demonstrated to be partly driven by genetic associations between DMI with these same energy sink traits, thereby having clear implications when comparing GWA studies on DMI to GWA studies on FE-like measures such as RFI. Objective 4a Data on Holstein (16,890), Brown Swiss (31,441), Simmental (25,845), and Alpine Grey (12,535) cows reared in northeastern Italy were used to assess the ability of milk components (fat, protein, casein, and lactose) and Fourier transform infrared (FTIR) spectral data to diagnose pregnancy. Pregnancy status was defined as whether a pregnancy was confirmed by a subsequent calving and no other subsequent inseminations within 90 d of the breeding of specific interest. Milk samples were analyzed for components and FTIR full-spectrum data using a MilkoScan FT+ 6000 (Foss Electric, Hillerod, Denmark). Pregnancy status was predicted using generalized linear models with fat, protein, lactose, casein, and individual FTIR spectral bands or wavelengths as predictors. We also fitted a generalized linear model as a simultaneous function of all wavelengths (1,060 wn) with a Bayesian variable selection model. Overall, the best prediction accuracies were obtained for the model that included the complete FTIR spectral data. We observed similar patterns across breeds with small differences in prediction accuracy. The highest CV-AUC value was obtained for Alpine Grey cows (CV-AUC = 0.645), whereas Brown Swiss and Simmental cows had similar performance (CV-AUC = 0.630 and 0.628, respectively), followed by Holsteins (CV-AUC = 0.607). For single-wavelength analyses, important peaks were detected at wn 2,973 to 2,872 cm(-1) where Fat-B (C-H stretch) is usually filtered, wn 1,773 cm(-1) where Fat-A (C=O stretch) is filtered, wn 1,546 cm(-1) where protein is filtered, wn 1,468 cm(-1) associated with urea and fat, wn 1,399 and 1,245 cm(-1) associated with acetone, and wn 1,025 to 1,013 cm(-1) where lactose is filtered. In conclusion, this research provides new insight into alternative strategies for pregnancy screening of dairy cows. Objective 4b The relationship of the estrous cycle to milk composition and milk physical properties was assessed on Holstein (n = 10,696), Brown Swiss (n = 20,501), Simmental (n = 17,837), and Alpine Grey (n = 8,595) cows reared in northeastern Italy. The first insemination after calving for each cow was chosen to be the day of estrus and insemination. Test days surrounding the insemination date (from 10 d before to 10 d after the day of the estrus) were selected and categorized in phases relative to estrus as diestrus high-progesterone, proestrus, estrus, metestrus, and diestrus increasing-progesterone phases. Milk components and physical properties were predicted on the basis of Fourier-transform infrared spectra of milk samples and were analyzed using a linear mixed model, which included the random effects of herd, the fixed classification effects of year-month, parity number, breed, estrous cycle phase, day nested within the estrous cycle phase, conception, partial regressions on linear and quadratic effects of days in milk nested within parity number, as well as the interactions between conception outcome with estrous cycle phase and breed with estrous cycle phase. Milk composition, particularly fat, protein, and lactose, showed clear differences among the estrous cycle phases. Fat increased by 0.14% from the diestrus high-progesterone to estrous phase, whereas protein concomitantly decreased by 0.03%. Lactose appeared to remain relatively constant over diestrus high-progesterone, rising 1 d before the day of estrus followed by a gradual reduction over the subsequent phases. Specific fatty acids were also affected across the estrous cycle phases: C14:0 and C16:0 decreased (−0.34 and −0.48%) from proestrus to estrus with a concomitant increase in C18:0 and C18:1 cis-9 (0.40 and 0.73%). Finally, urea, somatic cell score, freezing point, pH, and homogenization index were also affected indicating variation associated with the hormonal and behavioral changes of cows in standing estrus. Hence, the variation in milk profiles of cows showing estrus should potentially be taken into account for precision dairy farming management. Objective 4c. Nothing new to report this year.

Publications

  • Type: Journal Articles Status: Published Year Published: 2018 Citation: Toledo-Alvarado, H., A.I. Vazquez, G. de los Campos, R.J. Tempelman, G. Bittante, and A. Cecchinato. 2018. Diagnosing pregnancy status using infrared spectra and milk composition in dairy cows. Journal of Dairy Science 101:2496-2505.
  • Type: Journal Articles Status: Published Year Published: 2018 Citation: Zhou, Y., E.E. Connor, G.R. Wiggans, Y. Lu, R.J. Tempelman, S.G. Schroeder, H. Chen, and G.F. Liu. 2018. Genome-wide copy number variant analysis reveals variants associated with 10 diverse production traits in Holstein cattle. BMC Genomics 19:314.
  • Type: Journal Articles Status: Published Year Published: 2018 Citation: Nawaz, M.Y., F. Jimenez-Krassel, J.P. Steibel, Y. Lu, A. Baktula, N. Vukasinovic, L. Neuder, J.L.H. Ireland, J.J. Ireland and R.J. Tempelman. 2018. Genomic heritability and genome wide association analysis of anti-M�llerian hormone in Holstein dairy heifers. Journal of Dairy Science 101:8063-8075.
  • Type: Journal Articles Status: Published Year Published: 2018 Citation: Toledo-Alvarado, H., A.I. Vazquez, G. de los Campos, R.J. Tempelman, G. Gabai, A. Cecchinato and G. Bittante. 2018. Changes in milk characteristics and fatty acid profile during the estrous cycle in dairy cows. Journal of Dairy Science 101:9135-9153.
  • Type: Journal Articles Status: Accepted Year Published: 2018 Citation: Rovere, G., G. de los Campos, R. J. Tempelman, A. I. Vazquez, F. Miglior, F. Schenkel, A. Cecchinato, G. Bittante, H. Toledo-Alvarado, and A. Fleming. 2018. A landscape of the heritability of Fourier-Transform infrared spectral wavelengths of milk samples by parity and lactation stage in Holstein cows. Journal of Dairy Science (accepted).
  • Type: Conference Papers and Presentations Status: Published Year Published: 2018 Citation: Tempelman, R.J. 2018. Using genetic relationships to improve the design and analysis of animal science studies. Journal of Animal Science 96 (Suppl. 2):215-216.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2018 Citation: VanRaden, P., J. O'Connell, E. Connor, M. VandeHaar, R.J. Tempelman, and K. Weigel. 2018. Including feed intake data from U.S. Holsteins in genomic prediction. Page 125 in Proceedings of the 11th World Congress on Genetics Applied to Livestock Production, Auckland, New Zealand, February 11-16, 2018. Vol. Biology - Feed Intake and Efficiency 1.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2018 Citation: Rovere, G.A., G. de los Campos, A.I. Vazquez, A.L. Lock, L. Worden, and R.J. Tempelman. 2018. An assessment of different modelling strategies to predict milk fatty acid content using Fourier-transofrm infrared spectroscopy. Journal of Dairy Science 101 (Suppl.2):79-80.
  • Type: Journal Articles Status: Published Year Published: 2018 Citation: Lu, Y., M.J. Vandehaar, D.M. Spurlock, K.A. Weigel, L.E. Armentano, E.E. Connor, M. Coffey,R.F. Veerkamp, Y. de Haas, C.R. Staples, Z. Wang, M.D. Hanigan, and R.J. Tempelman. 2018. Genome wide association analyses based on a multiple trait approach for modeling feed efficiency. Journal of Dairy Science 101:3140-3154.


Progress 03/01/17 to 09/30/17

Outputs
Target Audience:Our target audience are fellow animal scientists and quantitative geneticists. Changes/Problems:No significant changes as of yet, particularly since this project only formally started 6 months ago. What opportunities for training and professional development has the project provided?Over the course of this short reporting period (03/01/2017-09/30/2017), two graduate students (Chunyu Chen and Yasir Nawaz) were trained under the project and successfully defended their PhD dissertation and Masters thesis, respectively. A newly arrived postdoctoral research associate Gabriel Rovere and a newly arrived Masters student Brooke Tate are also being trained under this project as well. How have the results been disseminated to communities of interest?The results have been presented in journal articles and at the national meetings of the American Dairy Science Association and the Italian Scientific Association of Animal Science What do you plan to do during the next reporting period to accomplish the goals?We hope to better develop the work for satisfying Objective #4 , especially the utility of Bayesian regression techniques to identify spectral data bands that are important predictors for health and reproduction outcomes in dairy cattle production. We also hope to use gas chromotography analyzed samples from Michigan dairy farms to determine which spectral regions are particularly important for which fatty acids and in turn draw associations between fatty acid composition and health and reproduction outcomes as influenced by energy balance. We also hope to finish and submit our work for formal inference on SNP markers responsible for genotype by environment interaction.

Impacts
What was accomplished under these goals? Objective 1. 1a) We have recently published a paper in the high impact journal Genetics which highlights the utility of hierarchical Bayesian modeling in conjunction with testing on genomic windows (i.e., joint tests on all genomic markers within a region rather than tests on individual markers themselves) to dramatically improve the sensitivity and specificity of genome wide association (GWA) inference relative to currently very popular strategies based on a standard normality assumption (Chen et al., 2017). That is, the hierarchical Bayesian approaches facilitate a wide range of distributional specification alternatives that more flexibly model genetic architecture for GWA studies. 1b) We currently have a paper under review (former graduate student Yongfang Lu) which utilizes both information on genotyped and non-genotyped animals for genome wide association inferences for feed efficiency and its component traits (dry matter intake, milk energy, and body weight). Additionally we have a manuscript in preparation (former graduate student Chunyu Chen) whereby we extend the work described above under 1a) to allow for non-genotyped animals having records within a more powerful hierarchical Bayesian modeling framework. Both papers illustrate how important it can be to include phenotypic information on non-genotyped animals to improve GWA inferences. Objective 2) 2a) We have recently published beef cattle related work with Brazilian collaborators (Mota et al, 2017) and dairy cattle related work with Wisconsin (Yao et al.,2017) collaborators. the interaction model provided a novel way to evaluate traits measured in multiple environments in which genetic heterogeneity may exist. The beef approach involves the use of reaction norm models to model genotype by environment interaction for tick resistance whereas the dairy study involved the use of marker by environment interaction specifications to allow estimation of environment-specific parameters and provided genomic predictions. Both analyses provided useful insights as to what genomic regions are particularly important drivers of genotype by environment interaction. 2b) We have work in progress to better identify and formally test for regions for genotype by environment interaction. We hope to publish this work over the following year. Objective 3. 3a) We have a paper under review that uses multiple trait GWA to distinguish the genetic architecture of feed efficiency from its component traits of body weight, dry matter intake and milk energy. The multiple trait model is particularly instructive as to what kinds of GWA relationships one should anticipate given the relationships between these traits. 3b) This is still a work in progress. Objective 4 4a) We have a paper under review that involves some of our international collaborators in Italy that demonstrate the utility of such approaches. 4b) This work has not yet been formally started. 4c) This work is just in its infancy.

Publications

  • Type: Journal Articles Status: Published Year Published: 2017 Citation: Yao, C., G. de los Campos, M. J. Vandehaar, D.M. Spurlock, L.E. Armentano, M. Coffey, Y. deHaas, R.F. Veerkamp, C.R. Staples, E.E. Connor, Z. Wang, M.D. Hanigan, R.J. Tempelman, and K.A. Weigel. 2017. Use of genotype � environment interaction model to accommodate genetic heterogeneity for residual feed intake, dry matter intake, net energy in milk, and metabolic body weight in dairy cattle. Journal of Dairy Science 100: 2007-2016. doi: 10.3168/jds.2016-11606.
  • Type: Journal Articles Status: Published Year Published: 2017 Citation: Mota, R.R., B.P. Sollero, P.S. Lopes, R.J. Tempelman, I. Aguilar, and F. F. Cardoso. 2017. Analyses of Reaction Norms Reveal New Chromosome Regions Associated with Tick Resistance in Cattle. Animal Jul 13:1-10. doi: 10.1017/S1751731117001562.
  • Type: Journal Articles Status: Published Year Published: 2017 Citation: Chen, C., J.P. Steibel, and R.J. Tempelman. 2017. Genome wide association analyses based on broadly different specifications for prior distributions, genomic windows, and estimation methods. Genetics 206:1791-1806. //doi: 10.1534/genetics.117.202259.
  • Type: Journal Articles Status: Published Year Published: 2017 Citation: Hardie, L.C., M. J. Vandehaar, R.J. Tempelman, K.A. Weigel, L.E. Armentano, G. Wiggans, R.F. Veerkamp, Y. deHaas, M. Coffey, E.E. Connor, M.D. Hanigan, C.R. Staples, Z. Wang, J.C.M Dekkers, and D.M. Spurlock 2017. The genetic and biological basis of feed efficiency in mid-lactation Holstein dairy cows. Journal of Dairy Science doi.org/10.3168/jds.2017-12604.
  • Type: Book Chapters Status: Published Year Published: 2017 Citation: Vandehaar, M.J., and R.J. Tempelman. 2017. Feeding and breeding to improve feed efficiency and sustainability. In: Large Dairy Herd Management, 3rd Edition. Beede, D.K. (Editor), American Dairy Science Association, Champaign, IL
  • Type: Conference Papers and Presentations Status: Published Year Published: 2017 Citation: Toledo-Alvarado, H., G. de los Campos, A.I. Vazquez, R.J. Tempelman. G. Bittante, and A. Cecchinato. 2017. Predictive ability of Fourier transform infrared spectroscopy and milk components to assess the pregnancy status of dairy cows. Italian Journal of Animal Science 16(Suppl. 1): 69.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2017 Citation: Connor, E.E., Y. Zhou, G.R. Wiggans, Y. Lu, R.J. Tempelman, S.G. Schoeder, H. Chen, and G. Liu. 2017. Genome-wide copy number variant analysis in Holstein cattle reveals variants associated with 10 production traits including residual feed intake and dry matter intake. Journal of Dairy Science 100 (Suppl.2):43.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2017 Citation: Nawaz, M.Y., F. Jimenez-Krassel, J.P. Steibel, Y. Lu, A. Baktula, N. Vukasinovic, S.K. Denise, L. Neuder, J.L.H. Ireland. J.J. Ireland, and R.J. Tempelman. Genome-wide association analysis and genomic heritability for anti-Mullerian hormone in Holstein dairy heifers. Journal of Dairy Science 100 (Suppl.2):132.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2017 Citation: Van Raden, P.M., J.R. Wright, E.E. Connor, M.J. Vandehaar, R.J. Tempelman, J.S. Liesman, L.E. Armentano, and K.A. Weigel. Preliminary genomic predictions of feed saved for 1.4 million Holsteins. Journal of Dairy Science 100 (Suppl.2):200-201.
  • Type: Conference Papers and Presentations Status: Published Year Published: 2017 Citation: Chen, C., K.A. Weigel, E.E. Connor, D.M. Spurlock, M.J. Vandehaar, C.R. Staples, and R.J. Tempelman. 2017. Bayesian whole-genome prediction and genome-wide association analysis with missing genotypes using variable selection. Journal of Dairy Science 100 (Suppl.2):412.