Award details

Sampling assisted parameter estimation

ReferenceBBS/E/R/02780629
Principal Investigator / Supervisor Professor Robin Thompson
Co-Investigators /
Co-Supervisors
Institution The Roslin Institute
DepartmentThe Roslin Institute Department
Funding typeResearch
Value (£) 68,995
StatusCompleted
TypeInstitute Project
Start date 01/04/1997
End date 31/03/2000
Duration36 months

Abstract

The basic objective will be to combine analytical and sampling based methods for even more efficient genetic parameter estimation in large, multivariate datasets. Previous work has shown that efficient estimation of genetic parameters is not possible for large commercial populations with multivariate data using exact analytical techniques. The problem is that the number of covariances to estimate grows in proportion to the square of the number of traits to be analysed, so with more than a few traits the number of parameters to estimate can be very large. We have shown that part of the iterative procedure can be speeded up dramatically by using specially constructed variables and predicting breeding values associated with these variables. The other part of the procedure can, in theory, be approximated by sampling generated variables and again predicting breeding values associated with the sampled variables. Various algorithms could be used to generate samples. The more complicated the sample generation schemes, the fewer the number of samples should be needed to give acceptable approximations. Indeed one can think of a continuum from an analytical approach to a purely sampling approach depending on how one initially estimates sampling variance associated with the breeding values. These different algorithms need evaluating in a variety of scenarios to give practical recommendations and allow Institute and industrial data to be analysed efficiently in both statistical and computational senses.

Summary

unavailable
Committee Closed Committee - Genes & Developmental Biology (GDB)
Research TopicsX – not assigned to a current Research Topic
Research PriorityX – Research Priority information not available
Research Initiative X - not in an Initiative
Funding SchemeX – not Funded via a specific Funding Scheme
terms and conditions of use (opens in new window)
export PDF file