test_predictor: Test accuracy of predictor on known phenotypes

Description Usage Arguments Value Examples

Description

This function takes the expression data input to build_predictor() and the coefficient estimates from build_predictor() for phenotype prediction. The known phenotypes are also input for comparison and asseessment of predictor accuracy.

Usage

1
2
test_predictor(inputdata = NULL, phenodata = NULL, phenotype = NULL,
  covariates = NULL, type = "factor", predictordata = NULL)

Arguments

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, binary, factor) type

predictordata

object output from build_predictor predictordata

Value

list of actual and predicted phenotype, and summarization of output

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
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")

## select 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)

## determine resubstitution error
## carry out prediction in training data set
predictions_test<-test_predictor(inputdata=inputdata ,phenodata=pheno,
	phenotype="sex", covariates=NULL,type="factor",predictordata=predictor)

ShanEllis/phenopredict documentation built on May 9, 2019, 1:24 p.m.