build_predictor: Uses output from filter_regions() to build the predictor
In ShanEllis/phenopredict: Predicting phenotypes from RNA-seq expression data

Description Usage Arguments Value Examples

This function takes the output from filter_regions() and builds the phenotype predictor to be used for phenotype prediction in a new data set. This function outputs the fits (coefficient estimates, 'coefEsts') from the model for the selectied regions as well as the rows selected ('trainingProbes') from the input data.

1 2	build_predictor(inputdata = NULL, phenodata = NULL, phenotype = NULL, covariates = NULL, type = NULL, numRegions = NULL)

`inputdata`	output from filter_regions() `inputdata`
`phenodata`	data set with phenotype information; samples in rows, variables in columns `phenodata`
`phenotype`	phenotype of interest `phenotype`
`covariates`	Which covariates to include in model `covariates`
`type`	The class of the phenotype of interest (numeric, factor) `type`
`numRegions`	The number of regions per class of variable of interest to pull out from each chromosome (default: 10) `numRegions`

An n x m data.frame of coefficient estimates and region indices for each of the regions included from filter_regions() along with regiondata and indices for trainingProbes

library('GenomicRanges')
library('dplyr')

## Make up some some region data
regions <- GRanges(seqnames = 'chr2', IRanges(
     start = c(28971710:28971712, 29555081:29555083, 29754982:29754984),
     end = c(29462417:29462419, 29923338:29923340, 29917714:29917716)))

## make up some expression data for 9 rows and 30 people
data(sysdata, package='phenopredict')
## includes R object 'cm'
exp= cm[1:length(regions),1:30]

## generate some phenotype information
sex = as.data.frame(rep(c("male","female"),each=15))
age = as.data.frame(sample(1:100,30))
pheno = dplyr::bind_cols(sex,age)
colnames(pheno) <- c("sex","age")

## filter regions to be used to build the predictor
inputdata <- filter_regions(expression=exp, regiondata=regions,
phenodata=pheno, phenotype="sex", covariates=NULL,type="factor",
numRegions=2)

## build phenotype predictor
predictor<-build_predictor(inputdata=inputdata ,phenodata=pheno,
	phenotype="sex", covariates=NULL,type="factor", numRegions=2)