binmod: binmod

In PAS: Polygenic Analysis System (PAS)

Description

This is the main function for bin model analysis.

Usage

binmod(x, y, map, beta0=NA, binsizelist=-1, full.search=FALSE, foldid=NA, ...)

Arguments

x	input matrix, of dimensions nobs*nvars; each row is a observation vector of an individual and each column is a genotypic indicator vector for a molecular marker.
y	a matrix of response variable (phenotypic observations), of dimensions nobs*1.
map	A data frame for linkage map or physical map.
beta0	Estimated SNP effects obtained by univariate analysis. By default, the glm function in R will be called by the binmod to calculate the estimates of effects.
binsizelist	A list of binsizes to be considered in the analysis. A default list will be generated if the option was ignored or an invalid list has been specified.
full.search	A logic indicator selecting search strategies. If FALSE was assigned, the binmod will complete the running as soon as the optimal binsize was found. Otherwise, analysis will be conducted for all binsizes on the list.
foldid	An optional vector of values between 1 and nfold identifying what fold each observation is in. If not supplied, a random vector is generated under nfold=10.
...	Other parameters need to be passed to glmnet/r and glm/r.

Details

The function invokes binmod analysis for genomic value prediction. The default settings are strongly suggested for new users.

Value

grid	information of all searched binsizes
grid$mselist	a 'data.frame': nbinsizes of 4 variables # A list of mean square errors
grid$mselist$binsize	size settings of the bins, eight in bp or cM.
grid$mselist$mse	mean square error
grid$mselist$mse_std	the standard deviation of MSEs
grid$mselist$nbin	number of bins under the binsize setting
grid$optbinsize	optimal binsize
grid$optid	order of the optimal binsize in the grid
optimal	result obtained under the optimal binsize
optimal$predict	phentypic values and its' predicted values under the optimal model.
optimal$predict$y	original phenotypic observations
optimal$predict$yp_cv	predictions by 10-fold cross-validation.
optimal$beta	estimated bin parameters
optimal$beta$beta	bin effect
optimal$beta$SSx	sum of square of bin indicator
optimal$beta$Se	residual error
optimal$beta$Sb	estimating error of bin effect
optimal$beta$Wald	Wald-test statistics
optimal$beta$LOD	LOD-test statistics
optimal$xbin	indicator matrix of the bins under the optimal binsize
optimal$map	'data.frame': of 5 variables: #bin map
optimal$map$chr	chromosome id
optimal$map$pos	bin position
optimal$map$pos_id	mean of the orders of markers in the bin
optimal$map$start_id	the order the first maker in a bin
optimal$map$end_id	the order the last maker in a bin
optimal$binsize	optimal binsize
optimal$cv	cross-validation results
optimal$cv$binsize	binsize
optimal$cv$nbin	number of bins under the binsize setting
optimal$cv$mse	mean squared error obtained from cross-validation
optimal$cv$r	Pearson's correlation coefficient obtained from cross-validation
snp	SNP information
snp$map	linkage map or physical map
snp$map$chr	chromosome id
snp$map$pos	marker position
snp$map$pos_id	marker order
snp$effect	single marker analysis result
snp$effect$beta	SNP effect
snp$effect$SSx	sum of square of genotypic indicator
snp$effect$Se	residual variance
snp$effect$Sb	estimating error of marker effect
snp$effect$Wald	Wald-test statistics
snp$effect$LOD	LOD test statistics
snp$mapinfo	a brief summary of the map
snp$mapinfo$chr	chromosome id
snp$mapinfo$start	the position of the first marker on the chromosome
snp$mapinfo$end	the position of the last marker on the chromosome
snp$mapinfo$length	length of the chromosome
snp$mapinfo$nmark	number of markers on the chromosome
snp$mapinfo$aver	average interval of the chromosome
snp$mapinfo$min.interval	the smallest interval size on the chromosome
cvfit	A cv.glmnet project. See manual of glmnet for details.

References

Zhiqiu Hu, Zhiquan Wang, and Shizhong Xu (2012) An infinitesimal model for quantitative trait genomic value prediction. PloS ONE

Examples

#load PAS library
library (PAS)
#load the demo data
data (beef)
#perform binmod analysis under the default settings.
binmod.result=binmod (x , y , map)
#plot binmod result
plot(binmod.result)
str(binmod.result)
#Output the predicted phenotypic values that was obtained
#by 10-fold cross validation .
predict(binmod.result)
#predict the phenotypic values for new individuals
x1=x[sample(1:NROW(x) , 20), ]
bin.pred.x1=predict(binmod.result, newx=x1)
str(bin.pred.x1)

PAS documentation built on May 2, 2019, 11:11 a.m.