Fast EM Algorithm for Genomic Selection
Author Information
Author(s): Shepherd Ross K, Meuwissen Theo HE, Woolliams John A
Primary Institution: CQUniversity, Rockhampton, Australia
Hypothesis
Can a fast EM algorithm improve genomic selection and complex trait prediction using genome-wide markers?
Conclusion
The emBayesB algorithm is a fast and accurate method for genomic selection that performs comparably to full Bayesian analyses but requires significantly less computational time.
Supporting Evidence
- emBayesB was found to be significantly faster than full Bayesian analyses, taking only minutes to converge.
- The accuracy of emBayesB was similar to that of Bayesian methods, with a correlation of 0.88 between GEBV and TBV.
- emBayesB detected all QTL with allele substitution effects greater than 0.18.
Takeaway
This study created a new computer program that helps predict traits in animals by looking at their genes, and it does this much faster than older methods.
Methodology
The study used an expectation-maximization (EM) algorithm called emBayesB to analyze SNP data and predict complex traits.
Limitations
The study does not provide standard errors routinely available from an EM algorithm, and the accuracy may vary with different genetic architectures.
Participant Demographics
The study involved 4665 individuals from a simulated population of livestock.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website