Differentially penalized regression to predict agronomic traits from metabolites and markers

Jane Ward, Mariann Rakszegi, Zoltán Bedő, Peter Shewry, Ian Mackay

Research output: Contribution to journalArticleResearchpeer-review

7 Citations (Scopus)

Abstract

Genomic prediction of agronomic traits as targets for selection in plant breeding programmes is increasingly common. The methods employed can also be applied to predict traits from other sources of covariates, such as metabolomics. However, prediction combining sets of covariates can be less accurate than using the best of the individual sets.

Results
We describe a method, termed Differentially Penalized Regression (DiPR), which uses standard ridge regression software to combine sets of covariates while applying independent penalties to each. In a dataset of wheat varieties, field traits are better predicted, on average, by seed metabolites than by genetic markers, but DiPR using both sets of predictors is best.

Conclusion
DiPR is a simple and accessible method of using existing software to combine multiple sets of covariates in trait prediction when there are more predictors than observations and the contribution to accuracy from each set differs.
Original languageEnglish
Number of pages7
JournalBMC Genetics
Volume16
Issue number19
DOIs
Publication statusFirst published - 26 Feb 2015
Externally publishedYes

Fingerprint

agronomic traits
metabolites
prediction
Software
Metabolomics
metabolomics
plant breeding
Genetic Markers
Triticum
Seeds
methodology
genomics
wheat
genetic markers
seeds
Plant Breeding
Datasets

Keywords

  • Genomic prediction
  • Metabolomics
  • wheat

Cite this

Ward, Jane ; Rakszegi, Mariann ; Bedő, Zoltán ; Shewry, Peter ; Mackay, Ian. / Differentially penalized regression to predict agronomic traits from metabolites and markers. In: BMC Genetics. 2015 ; Vol. 16, No. 19.
@article{4aac59cb2134452e9a761086c9f07390,
title = "Differentially penalized regression to predict agronomic traits from metabolites and markers",
abstract = "Genomic prediction of agronomic traits as targets for selection in plant breeding programmes is increasingly common. The methods employed can also be applied to predict traits from other sources of covariates, such as metabolomics. However, prediction combining sets of covariates can be less accurate than using the best of the individual sets.ResultsWe describe a method, termed Differentially Penalized Regression (DiPR), which uses standard ridge regression software to combine sets of covariates while applying independent penalties to each. In a dataset of wheat varieties, field traits are better predicted, on average, by seed metabolites than by genetic markers, but DiPR using both sets of predictors is best.ConclusionDiPR is a simple and accessible method of using existing software to combine multiple sets of covariates in trait prediction when there are more predictors than observations and the contribution to accuracy from each set differs.",
keywords = "Genomic prediction, Metabolomics, wheat",
author = "Jane Ward and Mariann Rakszegi and Zolt{\'a}n Bedő and Peter Shewry and Ian Mackay",
year = "2015",
month = "2",
day = "26",
doi = "10.1186/s12863-015-0169-0",
language = "English",
volume = "16",
journal = "BMC Genetics",
issn = "1471-2156",
publisher = "Springer Verlag",
number = "19",

}

Differentially penalized regression to predict agronomic traits from metabolites and markers. / Ward, Jane; Rakszegi, Mariann; Bedő, Zoltán; Shewry, Peter; Mackay, Ian.

In: BMC Genetics, Vol. 16, No. 19, 26.02.2015.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Differentially penalized regression to predict agronomic traits from metabolites and markers

AU - Ward, Jane

AU - Rakszegi, Mariann

AU - Bedő, Zoltán

AU - Shewry, Peter

AU - Mackay, Ian

PY - 2015/2/26

Y1 - 2015/2/26

N2 - Genomic prediction of agronomic traits as targets for selection in plant breeding programmes is increasingly common. The methods employed can also be applied to predict traits from other sources of covariates, such as metabolomics. However, prediction combining sets of covariates can be less accurate than using the best of the individual sets.ResultsWe describe a method, termed Differentially Penalized Regression (DiPR), which uses standard ridge regression software to combine sets of covariates while applying independent penalties to each. In a dataset of wheat varieties, field traits are better predicted, on average, by seed metabolites than by genetic markers, but DiPR using both sets of predictors is best.ConclusionDiPR is a simple and accessible method of using existing software to combine multiple sets of covariates in trait prediction when there are more predictors than observations and the contribution to accuracy from each set differs.

AB - Genomic prediction of agronomic traits as targets for selection in plant breeding programmes is increasingly common. The methods employed can also be applied to predict traits from other sources of covariates, such as metabolomics. However, prediction combining sets of covariates can be less accurate than using the best of the individual sets.ResultsWe describe a method, termed Differentially Penalized Regression (DiPR), which uses standard ridge regression software to combine sets of covariates while applying independent penalties to each. In a dataset of wheat varieties, field traits are better predicted, on average, by seed metabolites than by genetic markers, but DiPR using both sets of predictors is best.ConclusionDiPR is a simple and accessible method of using existing software to combine multiple sets of covariates in trait prediction when there are more predictors than observations and the contribution to accuracy from each set differs.

KW - Genomic prediction

KW - Metabolomics

KW - wheat

U2 - 10.1186/s12863-015-0169-0

DO - 10.1186/s12863-015-0169-0

M3 - Article

VL - 16

JO - BMC Genetics

JF - BMC Genetics

SN - 1471-2156

IS - 19

ER -