A postdoctoral position aiming to make it feasible to compute restricted maximum likelihood (REML) estimates of variance components in mixed models with many more traits, effects, and levels of effects than is currently feasible is available immediately at the U.S. Meat Animal Research Center. These improvements will be accomplished by combining theoretical developments that reimagine many aspects of the computational approaches currently used in variance component estimation (VCE).
Position Description:
The primary objective of the position is to investigate and report on computational aspects of REML variance component estimation. There are multiple potential manuscripts for which theory development, software implementation, and numerical confirmation are substantially developed and writing could commence immediately with a bit of collaborative expansion from the candidate and PI. Additional potential manuscripts which require more development to reach the writing stage could be undertaken as part of this position. There are many other opportunities for theory development, implementation in software, numerical confirmation, and publication in areas that have only been explored superficially. It is expected that the successful candidate will write several manuscripts from the previous categories. After that, the position could go in several different ways. Writing could continue on the queue of manuscripts to obtain as many publications as possible. Depending on interests and aptitudes, focus could shift to making substantial contributions to theoretical developments and/or software development.
Potential Impact:
It is expected that the successful candidate will contribute directly or indirectly to the development of an R package for analysis of mixed models and REML estimation of variance components. It is anticipated that this R package (and/or other software that implements the published computational approaches) will make it feasible to routinely estimate variance components in models including random effects with tens of millions of levels. The primary motivation for the project comes from quantitative genetics where genetic merit for several traits is routinely predicted for tens of millions of animals using variance components estimated by approximations with unknown properties because REML estimation is computationally infeasible. There are opportunities to improve these predictions of genetic merit by including data on many more correlated traits than is currently feasible. Although the objectives for this project are motivated by problems in genetics, many other disciplines use mixed models and software with improved estimation of variance components would benefit those disciplines as well.
Required Qualifications:
- Must be sufficiently familiar with REML and the mixed model equations to write manuscripts describing the theory and implementation of improved approaches to computing REML estimates
- Excellent communication (including written) skills and ability to work collaboratively
- Must be awarded a Ph.D. by May, 2024
- Must be a U.S. citizen or permanent resident
Preferred Qualifications:
- Ph.D. in statistics, genetics, computer science, mathematics, or similar field of study
- Interest and aptitude in development and application of statistical theory
- Experience in R programming and principles of software development
Compensation and Benefits:
This federal government position is funded for 2 years starting at the GS-11-1 level (salary for 2024 is $72,553). There is a roughly 3.3% step increase to GS-11-2 after 12 months of employment conditional on satisfactory performance and typically a cost of living increase each January 1. The benefits package includes excellent health and life insurance, 13 days per year of annual leave, 11 federal holidays, 13 days per year of sick leave and considerable flexibility for telework and work schedule. The successful candidate will accrue years of service for the Federal Employees Retirement System and is eligible for participation in the Thrift Savings Plan (TSP), in which the government will match employee contributions to the TSP account up to 5% of salary.
About the U.S. Meat Animal Research Center:
The U.S. Meat Animal Research Center (USMARC) has among the largest, if not the largest research populations of beef cattle, swine, and sheep in the world. It is located on 55 contiguous square miles near Clay Center, NE. USMARC conducts research to improve the productivity and end-product quality and safety of beef cattle, swine and sheep. USMARC is a very collaborative environment. Postdocs interact with leading scientists and other postdocs in a variety of disciplines including genetics and machine learning.
The U.S. Meat Animal Research Center is a facility of the U.S. Department of Agriculture’s (USDA) Agricultural Research Service, which provides the funding for this postdoctoral position.
To Apply:
Please submit a statement of interest, c.v., academic transcripts, and contact information for 2 references to Mark.Thallman@usda.gov. The position is open until filled.
For More Information:
Please contact Mark Thallman (Mark.Thallman@usda.gov; 402-762-4261) with any questions.
USDA/ARS is an equal opportunity provider and employer.