Progress 02/15/18 to 08/14/22
Outputs
Target Audience:
Nothing Reported
Changes/Problems:During the project, we did not experience any major issues or setbacks. Even during the pandemic, we were able to conduct our field experiments in collaboration with our industry partners without delays. Even though, it was not possible as planned to meet regularly in-person, conduct face-to-face training workshops, as well as conduct field and nursery visits, we kept a regular virtual meeting schedule. What opportunities for training and professional development has the project provided?
Nothing Reported
How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?
Nothing Reported
Impacts
What was accomplished under these goals?
Objectives 1 and 2 - The 375 RILs were derived from biparental crosses between two Stiff-Stalk inbreds and four non-Stiff Stalk inbreds in previous, partially NIFA-funded projects. All 375 recombinant inbreds (RILs) were genotyped by Corteva, using the MaizeSNP50 chip assay. In collaboration with NRGene, we obtained additional sequencing information for these RILs. Together, these data allowed us to locate with high accuracy recombination events, copy number variation, and present/absent variation in our experimental RILs' genome. Objective 3 - In the 2018 and 2019 maize breeding nurseries at the University of Illinois and the University of Minnesota, 332 recombinant inbred lines (RILs) were crossed to develop 400 experimental hybrids. The hybrids were evaluated for their agronomic performance in eight locations in Illinois, Minnesota, and Wisconsin in 2019. These experiments were repeated in 2020. In addition to the yield trials conducted at the agricultural experimental stations of the University of Illinois and the University of Minnesota, the following companies evaluated the experimental hybrids in replicated yield trials in 2019 and 2020: Corteva (total number of environments: 4), Bayer (2), Syngenta (2), Becks (4 locations), and AgReliant (1). The 2020 season was characterized by a cold and wet spring, causing planting delays and heterogenous field stands at most locations. Dramatic weather events during the growing season, including the Derecho (8/2020) and a record hailstorm in East Central Illinois (7/2020), caused severe damage. Out of eight locations, six could be harvested, and four provide quality phenotypic data. Several phenotypic traits were measured, including yield, plant height, ear height, and kernel test weight and moisture. The amount of variation in General (GCA) and specific combining ability (SCA) effects explained by structural variants was quantified using a two-stage mixed linear model approach. In the first stage, BLUEs were calculated on a single environment basis for each trait. In the second stage, parent relationship matrices and hybrid dominance matrix were included in a multi-environment model. The genotypic markers used to calculate the dominance matrices were split into five categories: SNPs+SVs, only SNPs, only SVs, SNPs in LD with SVs, and SNPs not in LD with SVs. Here SV represents structural variants, i.e., the total PAV and CNV variants, and SNP a single nucleotide polymorphism. The second stage resulted in GCA and SCA variances for each trait and marker type. Except for grain moisture for the "Parent A" hybrid group and yield for the "Parent B" hybrid group, GCA variances were greatest in SV models for all traits. GCA variances were generally smaller than SCA variances. Objective 4 - We used the genomic information of all RILs and the 2019 phenotypic data of 400 hybrid progeny to predict the performance of all possible hybrids (N > 70K). We chose 100 untested hybrids based on these predictions to validate our preliminary prediction model. Seeds for these untested hybrids were produced in the 2020 breeding nurseries at the Universities of Illinois and Minnesota. In 2021, we evaluated this set of 100 untested hybrids ("Validation Experiment") for their agronomic performance in a total of six environments in Illinois (N=4), Minnesota (N=2), and Iowa (N=1). Our industry partners AgReliant and Becks Hybrids provided three locations. Breeders commonly use genetic markers to predict the performance of untested individuals as a way to improve the efficiency of breeding programs. These genomic prediction models have almost exclusively used SNPs as their source of genetic information, even though other types of markers exist, such as SVs. Given that not all SVs are in linkage disequilibrium to SNPs and that they are associated with environmental adaptation, SVs have the potential to bring additional information to multi-environment prediction models that are not captured by SNPs alone. Here, we evaluated different marker types (SNPs and/or SVs) on prediction accuracy across a range of genetic architectures for simulated traits across multiple environments. Our results show that SVs can improve prediction accuracy by up to 19%, but it is highly dependent on the genetic architecture of the trait. Differences in prediction accuracy across marker types were more pronounced for traits with high heritability, high number of QTLs, and SVs as causative variants. In these scenarios, using SV markers resulted in better prediction accuracies than SNP markers, especially when predicting untested genotypes across environments, likely due to more predictors being in linkage disequilibrium with causative variants. The simulations revealed little impact of different effect sizes between SNPs and SVs as causative variants on prediction accuracy. This study demonstrates the importance of knowing the genetic architecture of a trait in deciding what markers and marker types to use in large scale genomic prediction modeling in a breeding program. Building Center of Excellence - We communicated in biweekly virtual meetings during the project's duration and continued to do so after the project ended. We use our Center of Excellence to continue our collaboration by publishing more manuscripts based on the findings of this project and creating new research ideas and subsequent research proposals. We added new faculty to our center - Dr. Sam Fernandes contributed to our project as a postdoctoral scientist in Dr. AE Lipka's research group. Dr. Fernandes recently took on a tenured faculty position at the University of Arkansas. He significantly increases UA's research capability in genomic data science and agricultural statistics. We trained two award winning graduate students. Rafa Della Coletta (UMN) was awarded the MnDRIVE Fellowship (Minnesota's Discovery, Research, and InnoVation Economy). Sharon E Liese was awarded the IllinoisCorn Marketing Board fellowship.
Publications
- Type:
Other
Status:
Published
Year Published:
2022
Citation:
Liese SE, R Della Coletta, SB Fernandes, CN Hirsch, AE Lipka, MA Mikel and MO Bohn. 2022. P62-The impact of structural variation on heterosis and combining ability in maize (Submitted by Sharon Liese). 64th Annual Maize Genetics Meeting. https://documents.maizegdb.org/maizemeeting/abstracts/2022Program.pdf.
- Type:
Other
Status:
Published
Year Published:
2022
Citation:
Della Coletta R, SB Fernandes, PJ Monnahan and CN Hirsch. 2022. P67-Understanding the relationship between genetic architecture and genetic marker selection for improved genomic prediction accuracy. 64th Annual Maize Genetics Meeting. https://documents.maizegdb.org/maizemeeting/abstracts/2022Program.pdf.
- Type:
Journal Articles
Status:
Under Review
Year Published:
2023
Citation:
Della Coletta R, SE Liese, SB Fernandes, MA Mikel, MO Bohn, AE Lipka and CN Hirsch. 2023. Linking genetic and environmental factors through marker effect networks to understand trait plasticity. bioRxiv. doi: https://doi.org/10.1101/2023.01.19.524532.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2021
Citation:
Della Coletta, R, Fernandes, S, Monnahan, PJ, O'Connor, CJ, Lipka, AE, Bohn, MO, Mikel, MA and Hirsch, CN. 2021. Leveraging structural variant information in GxE genomic prediction models [Abstract]. ASA, CSSA, SSSA International Annual Meeting, Salt Lake City, UT. https://scisoc.confex.com/scisoc/2021am/meetingapp.cgi/Paper/137615.
- Type:
Other
Status:
Published
Year Published:
2021
Citation:
Della Coletta R, SB Fernandes, PJ Monnahan, C O'Connor, AE Lipka, MO Bohn, MA Mikel and CN Hirsch. 2021. P73-Leveraging structural variant information in GxE genomic prediction. 63th Annual Maize Genetics Meeting, 2nd virtual Meeting. https://documents.maizegdb.org/maizemeeting/abstracts/2021Program.pdf.
- Type:
Journal Articles
Status:
Submitted
Year Published:
2023
Citation:
Della Coletta R, SB Fernandes, PJ Monnahan, MA Mikel, MO Bohn, AE Lipka and CN Hirsch. 2023. Importance of genetic architecture in marker selection decisions for genomic prediction. Theoretical and Applied Genetics (Submitted).
|
Progress 02/15/21 to 02/14/22
Outputs
Target Audience:
Nothing Reported
Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?
Nothing Reported
How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?The project is on target with all proposed research activities. We are in the process of finalizing and publishing all bioinformatics analyses (Objective 4). Project graduate students Rafa Della Coletta (University of Minnesota) and Sharon Liese (University of Illinois) will graduate at the end of the Spring Semester 2022.
Impacts
What was accomplished under these goals?
Objectives 1 and 2 - The 375 RILs were derived from biparental crosses between two Stiff-Stalk inbreds and four non-Stiff Stalk inbreds in previous, partially NIFA-funded projects. All 375 recombinant inbreds (RILs) were genotyped by Corteva, using the MaizeSNP50 chip assay. In collaboration with NRGene, we obtained additional sequencing information for these RILs. Together, these data allowed us to locate with high accuracy recombination events, copy number variation, and present/absent variation in our experimental RILs' genome. Objective 3 - In the 2018 and 2019 maize breeding nurseries at the University of Illinois and the University of Minnesota, 332 recombinant inbred lines (RILs) were crossed to develop 400 experimental hybrids. The hybrids were evaluated for their agronomic performance in eight locations in Illinois, Minnesota, and Wisconsin in 2019. These experiments were repeated in 2020. In addition to the yield trials conducted at the agricultural experimental stations of the University of Illinois and the University of Minnesota, the following companies evaluated the experimental hybrids in replicated yield trials in 2019 and 2020: Corteva (total number of environments: 4), Bayer (2), Syngenta (2), Becks (4 locations), and AgReliant (1). The 2020 season was characterized by a cold and wet spring, causing planting delays and heterogenous field stands at most locations. Dramatic weather events during the growing season, including the Derecho (8/2020) and a record hail storm in East Central Illinois (7/2020), caused severe damage. Out of eight locations, six could be harvested, and four provide quality phenotypic data. Objective 4 - We used the genomic information of all RILs and the 2019 phenotypic data of 400 hybrid progeny to predict the performance of all possible hybrids (N > 70K). We chose 100 untested hybrids based on these predictions to validate our preliminary prediction model. Seeds for these untested hybrids were produced in the 2020 breeding nurseries at the University of Illinois and the University of Minnesota. In 2021, we evaluated this set of 100 untested hybrids ("Validation Experiment") for their agronomic performance in a total of sevenenvironments in Illinois (N=4), Minnesota (N=2), and Iowa (N=1). Our industry partners AgReliant and Becks Hybrids provided three locations. Project findings were presented at the 2021 Maize Genetics Conference (virtual conference) and the 2021 ASA, CSSA, SSSA International Annual Meeting in Salt Lake City, Utah. Simulation experiments were conducted to test how the composition of causative variants (SNPs, structural variants, or both) and their interaction with environments impact the prediction accuracy of inbreds and single cross hybrids for traits with different genetic architectures (i.e., number of genes, additive and dominant effects, genotype-by-environment interactions). We currently use the results of these simulation experiments to gauge the impact of structural genomic variation on heterosis, general and specific combining abilities, and prediction accuracy in maize using our experimental data collected in 2019, 2020, and 2021.
Publications
|
Progress 02/15/20 to 02/14/21
Outputs
Target Audience:
Nothing Reported
Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?
Nothing Reported
How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?The project is on target with all proposed research activities. The validation experiment will be conducted at the Universityof Illinois (2 locations) and the University of Minnesota (2 locations) and three additional experimental sites provided by industry partners. The validation experiment consists of 100 untested hybrids (see this report - Objective 4). All bioinformatics analyses (Objective 4) will be finalized and published.
Impacts
What was accomplished under these goals?
Objective 2 - The 375 RILs were derived from biparental crosses between two Stiff-Stalk inbreds and four non-Stiff Stalk inbreds in previous, partially NIFA-funded projects. All 375 recombinant inbreds (RILs) were genotyped by Corteva, using the MaizeSNP50 chip assay. In collaboration with NRGene, we obtained additional sequencing information for these RILs. Together, these data allowed us to locate with high accuracy recombination events, copy number variation, and present/absent variation in our experimental RILs' genome. Objective 3 - In the 2018 and 2019 maize breeding nurseries at the University of Illinois and the University of Minnesota, 332 recombinant inbred lines (RILs) were crossed to develop 400 experimental hybrids. The hybrids were evaluated for their agronomic performance in eight locations in Illinois, Minnesota, and Wisconsin in 2019. These experiments were repeated in 2020. In addition to the yield trials conducted at the agricultural experimental stations of the University of Illinois and the University of Minnesota, the following companies evaluated the experimental hybrids in replicated yield trials in 2019 and 2020: Corteva (total number of environments: 4), Bayer (2), Syngenta (2), Becks (4 locations), and AgReliant (1). The 2020 season was characterized by a cold and wet spring, causing planting delays and heterogenous field stands at most locations. Dramatic weather events during the growing season, including the Derecho (8/2020) and a record hail storm in East Central Illinois (7/2021), caused severe damage. Out of eight locations, six could be harvested, and four provide quality phenotypic data. Objective 4 - We used the genomic information of all RILs and the 2019 phenotypic data of 400 hybrid progeny to predict the performanceof all possible hybrids (N > 70K). Based on these predictions, we selected 100 untested hybrids to validate our preliminary prediction model. Seeds for these untested hybrids were produced in our 2020 breeding nurseries at the Universityof Illinois and the University ofMinnesota for evaluation experiments in 2021. We conducted simulation experiments to determine how the type of causative variants (SNPs, Structural Variants (SVs), or both) impact the prediction accuracy of inbreds and single cross hybrids for traits with different genetic architectures (i.e., number of genes, additive and dominant gene effects, genotype-by-environment interactions). Each genetic architecture was simulated for a single and multiple environments using the R package simple PHENOTYPES developed in the group of Co-PI Lipka. The results of these simulation experiments already provided useful information about the impact of SVs on prediction accuracy that will guide the analysis of our experimental data collected in 2019 and 2020.
Publications
|
Progress 02/15/19 to 02/14/20
Outputs
Target Audience:
Nothing Reported
Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?During the summer of 2019, we provided graduate students with a project-internal training workshop on genomic selection at the University of Illinois. How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?The project is on target with all proposed research activities.2020 will be the second project year withfield experiments, i.e., 400 experimental hybrids will beevaluated for their agronomic performance at eight locations throughout the U.S. corn belt (Objective 3). Inthe first half of 2020, we will continue withthe development of our simulation pipeline and we willmap parental copy number variants to their RIL offspring (Objectives 1 and 2). Relatingthis genetic information with the2019 performance data will provide us with the first insights into how SVs impact agronomic performance in maize (Objective 4). We will produce seed for 100 untested hybrids to validate prediction models developed in 2019. During the summer of 2020, we will provide graduate students with a project-internal training workshop on bioinformatics at the University of Minnesota.
Impacts
What was accomplished under these goals?
Objective 2 - The 375 RILs were derived from biparental crosses between two Stiff-Stalk inbreds and four non-Stiff Stalk inbreds in previous, partially NIFA-funded projects. All 375 recombinant inbreds (RILs) were genotyped by Corteva applying the MaizeSNP50 chip assay. We obtained additional sequencing information for these RILs in collaboration with NRGene. Together these data will allow us to locate recombination events in the maize genome with high accuracy. Currently, we have simulated quantitative traits of inbred lines using genotypic information from a population of 525 RILs derived from diallel crosses of seven maize inbred parental lines for a single environment. The 375 RILs used in our experiment are a subset of this 525 RIL set. We simulated traits with additive effects assuming different heritabilities (0.2, 0.5, or 0.9), the number of loci (3, 25 or 75), as well as types of causative variants (SNPs, Structural Variants (SVs), or both) controlling the trait. Each genetic architecture was simulated for a single environment using the R package simple phenotypesdeveloped in the lab of Co-PI Dr. Lipka. We tested genomic prediction models using SNPs or SVs or both as markers applying RR-BLUP, and compared prediction accuracy for each genetic architecture-marker combination using a 5-fold cross-validation scheme. We also tested a different number of markers as predictors in RR-BLUP (all polymorphic markers, 1,000 random polymorphic markers, or 50 random polymorphic markers). Overall, our preliminary results indicate that the number of loci does not seem to have a significant impact on prediction accuracy, while prediction accuracy increases as heritability increases. In prediction models with all markers available, we observe that when a trait is controlled only by SVs, using SVs as predictors results in higher accuracy than using SNPs except when heritability is very low. Similarly, when a trait is exclusively controlled by SNPs, SVs are not good predictors. Using a set of 1,000 random markers for genomic prediction yielded very similar results as if using all markers, but results were inconclusive when using only 50 markers. We are working on expanding our pipeline to simulate phenotypes of RILs for multiple environments, and simulating traits for maize hybrids derived from the RIL population described above by adding dominance effects. Objective 3 - In the 2018 maize breeding nurseries at the University of Illinois and the University of Minnesota, 332 recombinant inbred lines (RILs) were crossed to develop 400 experimental hybrids. The goal was to produce sufficient seed of each hybrid for tests in at least 20 environments. We missed this goal for 70 hybrids for which the seed supply was adequate only for yield trials in 2019. Seed production for these hybrids was repeated successfully in the 2019 nurseries at the Universities of Illinois and Minnesota. The seed of all 400 experimental hybrids was treated and packaged at the University of Illinois and send to each company. The hybrids were evaluated for their agronomic performance in eight locations in Illinois, Minnesota, and Wisconsin. In addition to the yield trials conducted at the agricultural experimental stations of the University of Illinois and the University of Minnesota, the following companies evaluated the experimental hybrids in replicated yield trials: Corteva (2 locations), Bayer (1 location), Syngenta (1 location), and Becks (2 locations). AgReliant was not able to provide yield trial opportunities in 2019, but will do so in 2020. Even though the 2019 growing season was characterized by severely delayed planting at all locations, we obtained reliable performance data from six out of the eight locations. Objective 4 - We used the genomic information of all RILs and the 2019 phenotypic data of 400 hybrid progeny to predict the performs of all possible hybrids (N > 70K). Based on these predictions, we selected 100 untested hybrids to validate our preliminary prediction model. Seeds of these untested hybrids will be produced in our 2020 breeding nurseries at the Universities of Illinois and Minnesota for evaluations experiments in 2021.
Publications
|
Progress 02/15/18 to 02/14/19
Outputs
Target Audience:
Nothing Reported
Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?
Nothing Reported
How have the results been disseminated to communities of interest?We have not yet disseminated results as the first year was dedicated to hybrid seedproduction and genotypingof parental recombinant inbred lines. What do you plan to do during the next reporting period to accomplish the goals?The project is on target with all proposed research activities.2019 will be the first project year withfield experiments, i.e., 400 experimental hybrids will beevaluated for their agronomic performance at eight locations throughout the U.S. corn belt (Objective 3). Inthe first half of 2019, we will focus onObjectives 1 and 2, and map parental copy number variants to their RIL offspring. Relatingthis genetic information with theperformance data will provide us with the first insights of how genome structure variants impact agronomic performance in maize (Objective 4). During the summer of 2019, we will provide graduate students with a project-internal training workshop on genomic selection at the University of Illinois and bioinformatics at the University of Minnesota.
Impacts
What was accomplished under these goals?
Objective 2 - All 375 recombinant inbreds (RILs) were genotyped by DOW AgroSciences (now Corteva) applying the MaizeSNP50 chip assay. These data will be used for further bioinformatics analysis. Objective 3 - In 2018 inmaize breeding nurseries at the University of Illinois and the University of Minnesota, 375 recombinant inbred lines (RILs) were crossed to develop 400 experimental hybrids. The 375 RILs were derived from biparental crosses between two Stiff-Stalk inbreds and four non-Stiff Stalk inbreds in previous, partially NIFA-funded projects. The crossing nursery was replicated at both universities to ensure sufficient production of hybrid seed for large scale and multi-location field experiments in 2019. Seed produced at the University of Minnesota was shipped to the University of Illinois for inventory and storage. We were successful in producing seed numbers for 330 experimental hybrids sufficient for all planned project field evaluations in 2019 and 2020. For the remaining 70 experimental hybrids, sufficient seed is available for 2019. However, a second round of seed production for these hybrids in 2019 is necessary to ensure ample seed for experiments in 2020. The following companies agreed to evaluate the experimental hybrids in the growing season of 2019: Corteva (2 locations), Bayer (1 location), Syngenta (1 location), and Becks (2 locations). AgReliant will not be able to provide yield trial opportunities in 2019 but will do so in 2020 (2 locations). The seed was treated, packaged, and send to each company in time for testing in 2019. Rafael Della Coletta was recruited as a graduate student at the University of Minnesota. At the University of Illinois, Sharon Folk was recruited. Shewill start her graduate studies in the summer of 2019.
Publications
|
|